- Email: [email protected]
Implications for forest management of the EU Water Framework Directive's stream water quality requirements — A modeling approach
Forest Policy and Economics 13 (2011) 284–291
Contents lists available at ScienceDirect
Forest Policy and Economics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / f o r p o l
Implications for forest management of the EU Water Framework Directive's stream water quality requirements — A modeling approach Ljusk Ola Eriksson a,⁎, Stefan Löfgren b, Karin Öhman a a b
Swedish University of Agricultural Sciences, Department of Forest Resource Management, S-901 83 Umeå, Sweden Swedish University of Agricultural Sciences, Department of Aquatic Sciences and Assessment, P.O. Box 7050, S-750 07, Uppsala, Sweden
a r t i c l e
i n f o
Article history: Received 17 June 2010 Received in revised form 24 December 2010 Accepted 3 February 2011 Available online 15 March 2011 Keywords: Long-term planning Linear programming EU Water Framework Directive Nutrients Methyl mercury Dissolved organic carbon
a b s t r a c t The EU Water Framework Directive (WFD) stipulates that measures should be taken to ensure that all lakes and streams in the EU have good ecological and chemical status, comparable to that of waters unaffected by human activities. This has profound potential implications for forestry, since operations such as harvesting and fertilization tend to reduce the quality of ground and stream water in affected catchments. The aim of this study is to assess the implications for forestry of limiting the concentrations of chemical substances reaching lakes and streams. A forest planning model with a horizon of 100 years that includes requirements regarding water concentrations of nitrogen (N), phosphorus (P), methyl mercury (MeHg) and dissolved organic carbon (DOC) was applied to three intensively-studied sub-catchments at Balån in the boreal part of northern Sweden. Limiting maximum increases in concentrations of these substances to 10% above reference values resulted in an economic loss of ca. 20% or 35%, depending on whether the limits were applied to the whole area or to each sub-catchment individually. The results were also highly dependent on the assumptions, especially regarding the flux of MeHg. The results should be interpreted with caution as there are still major uncertainties concerning the cause and effect relationships. We also need to consider additional aspects to those addressed here, such as acidification, erosion and the biological effects of the operations. © 2011 Elsevier B.V. All rights reserved.
1. Introduction The EU Water Framework Directive (WFD, 2000/60/EC) has set an objective to ensure that by 2015 the ecological and chemical status of all lakes and streams in Europe should be ‘good’, defined in terms of comparability to the water quality in pristine areas, unaffected by human activities. This is likely to have consequences for forestry operations at both stand and catchment levels, since hydrological, biological and chemical processes are all influenced by forestry activities (in addition to forest yields). Most importantly, in this context, operations such as forest harvesting and fertilization tend to reduce the water quality of streams and lakes receiving water from the forests, relative to those receiving water from non-managed forests (Piirainen et al., 2007; Kreutzweiser et al., 2008). However, the magnitude of these effects is clearly related to the proportion of the catchment affected by forestry operations and how these operations are performed (Ring et al., 2008). In Sweden, implementation of the WFD is led by five water authorities, which are authorized to apply management plans to enhance water quality, including plans to address effects of forestry, ⁎ Corresponding author at: Department of forest resource management, SLU, S-901 83 UMEÅ, Sweden. Tel.: + 46 706440004; fax: + 46 90778116. E-mail addresses: [email protected] (L.O. Eriksson), [email protected] (S. Löfgren), [email protected] (K. Öhman). 1389-9341/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.forpol.2011.02.002
where necessary to meet WFD requirements. A key task for these authorities is to assess the ecological status of waters. For this purpose, the Swedish Environmental Protection Agency (SEPA) has developed a classification system, which includes biological, chemical and hydro-morphological variables (SEPA, 2007, 2008). Hence, knowledge of the effects of forestry activities on values of these parameters is essential for understanding the adaptations forestry will need to make to meet the stipulations of the WFD as interpreted by the Swedish water authorities. Based on the SEPA system, the water authorities classify the ecological status as less than good in 39% and 44% of Swedish lakes and streams, respectively (Water Authorities, 2010). One reason for this is that the chemical status of the water in many forest streams tends to be close to the border between good and moderate classes (Löfgren et al., 2009a), which is the threshold for remedial actions according to the WFD. In many of these limnic systems, forestry is the major human influence, and a small increase in (for instance) phosphorus concentration caused by forestry might induce a shift from good to moderate class, and thus initiate a requirement for remedial actions. Another contributory factor, in some cases, may be incorrect classification based on the biological indices (Löfgren et al. op. cit., see below). Guidelines for mitigating the adverse effects of forest operations on water quality have been issued by the Swedish Forest Agency (2000) and the Swedish Forest Stewardship Council (1998). Ways in which forestry could adapt to meet water quality requirements have
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
not been thoroughly investigated either. Some “rules of thumb”, such as not allowing more than 30% of a catchment to be harvested (Ring et al., 2008), have been proposed, but it is not clear if they would be either optimal or sufficient. Further, one of few relevant studies under boreal conditions (Öhman et al., 2009) indicated that ensuring that the concentration of dissolved organic carbon (DOC) in streams and lakes fed by groundwater to b10% above the reference value could incur economic losses (relative to current practices) of 5–30%, depending on the assumed magnitude of forestry effects on water quality variables. A more comprehensive understanding of the relation between water quality and forestry than Öhman et al. (2009) is attempted here by allowing several modeled water quality variables to be simultaneously affected by forest operations. However, limitations in knowledge of the relationships between forestry activities and water quality parameters preclude analysis of many variables. Biological effects are inherently difficult to assess, not the least because the lack of casual relationship between organism status and water quality. For example, it is not possible to distinguish between natural acidity and acidification caused by human activity from the status of benthic fauna and fish (Löfgren et al., 2009a). Hydromorphological disturbances in the study area are basically of historical origin and less related to present forestry (Löfgren et al., 2009a, see below). Hence, the water quality variables that will be analyzed in this report are limited to water chemistry. In the study area, it is possible to estimate initial chemical parameters of the streams, and to make reasonable quantified assessments of the responses of these parameters to forest activities. The focus is on forestry as a large scale, long-term activity, i.e. the perspective is strategic. Hence, some effects of forestry activities on water quality, such as machinery crossing streams during harvesting operations, are not included as they are dependent on specific operational plans. These aspects, important as they may be (Nisbet, 2001; Kreutzweiser et al., 2008; Ring et al., 2008), are assumed to be accounted for in consistency with the values assigned by the water quality models used in this study. The strategic perspective also reduces the importance of hydro-morphological parameters as they are related more strongly to measures intended to improve the water quality (restoration of rafting channels for example), than normal forest operations. However it should be noted that there are exceptions, like for instance the abundance of large woody debris in streams caused by normal forest operations. The aim of this study was to assess, in a strategic setting, the potential implications for forestry of limiting the concentrations of selected chemical substances in surface waters below certain threshold values over time. To do this we used a traditional forest planning model aimed at maximizing the net present value (NPV) from future harvest activities subject to traditional forest constraints and requirements regarding water quality. The water quality variables are concentrations of nitrogen (N), phosphorus (P), methyl mercury (MeHg) and dissolved organic carbon (DOC). Model outputs pertain to financial values, forest management activities and concentrations of the substances mentioned above during the next 100 years. The analysis focuses on three intensively studied sub-catchment areas at Balån in the boreal part of northern Sweden (Löfgren et al., 2009b). Further, since water quality (like most ecological phenomena) is dependent on spatial scale, the relationships between forestry activities and water quality are assessed at two geographical scales: sub-catchment and whole catchment.
285
20°15′ E, Fig. 1). The area was chosen since it has been surveyed since 2004 and much of the data needed were available from the area (see articles in the special issue on Balån in Ambio Vol. 38, No. 7, Nov. 2009). The area consists of a total of 66 stands distributed on three sub-catchments I, II and III (Fig. 1). Areas I and II contain stands that on average are considerably older than those in area III, whereas the three stands together have a fairly even age distribution. The catchment is dominated by Scots pine (Pinus sylvestris) and Norway spruce (Picea abies) with limited occurrence of other species, mostly birch (Betula spp.) (Table 1). The forest data for the area were estimated by the k nearest neighbor method (kNN) (Reese et al., 2003), with stands delineated with the algorithm developed by Hagner (1990). Note that the description of the area given by Löfgren et al. (2009b) is based on forest data obtained from the land owner and a stump survey, not completely in agree with the kNN data. This is reflected, for instance, in the site index, which is lower on average in the kNN data. 2.2. Water quality indicators Changes in concentrations of relevant substances can be used as indicators of the effects of specific forest operations, and can be calculated by dividing their respective fluxes by the water runoff before and after the operations. For example, if the N flux at base level is 1.5 kg ha−1 y−1 and the annual water runoff is 400 mm y−1 (corresponding to an annual runoff volume of 0.4 m3 water from each square meter of the catchment), the N concentration in the runoff water will be 0.375 mg N L−1 [(1.5 * 106 mg N/104 m2)/ 0.4 m = 375 mg N m−3 = 0.375 mg N L−1]. Additionally, if 10% of the catchment area is clear-felled, leaching 3.0 kg N ha−1 y−1 and runoff 560 mm water y−1, the concentration in water from the clear-felled area will be (3.0 * 106 mg N/104 m2)/0.56 m = 0.536 mg N L−1] and from the entire catchment 0.391 mg N L−1 [=0.375 * 0.9 + 0.536 * 0.1]. This procedure has been followed here, since in most scientific literature the excess loss of a substance coupled to a certain forest operation is usually expressed as annual flux per hectare. Since the functions for growth and yields used in the study operate with a time step of 5 years, the description of the effects on water quality is also given as averages for 5-year periods. Further, in the study it is assumed that the consequences of a forest activity last for at most two 5-year periods, i.e. 10 years. 2.3. Establishing model parameters The model parameters used in this study are presented in Table 2, which summarizes the average effect per year of various forest operations on selected substances, as well as the base levels in non-managed forests. The rationale for setting these parameters is given below. The increased flux values for all operations except fertilization are based on the rate of estimated increase after a 1 ha harvest (clear-felling). In the case of fertilization, the effect is from fertilizing 1 ha. 2.3.1. Water runoff (W) The base level runoff is set to 400 mm y−1, following data from Balån presented by Löfgren et al. (2009b; Table 3). The increase from final felling is estimated with the model presented by Öhman et al. (2009; Table 1, Case 1a), resulting in an increase in the first 5-year period of 40% and in the second 5-year period of 25% over the base level. Increased water flow following harvest has been assumed when calculating the post-harvest (1st and 2nd periods) concentrations for all substances.
2. Materials and methods 2.1. Study area We set parameter values for our model based on data derived from the Balån study area of north-eastern Sweden (Löfgren et al., 2009a, b), situated ca. 60 km from the coast in the county of Västerbotten (63°49′ N,
2.3.2. Nitrogen (N) and clear-felling The data given by Löfgren and Olsson (1990) and the conclusions in Löfgren (2007) indicate that clear-felling increases N fluxes from about 1.5 to 3.0 kg N ha−1 y−1 on average over an entire period of 10 years across the Bothnian Bay region. These values describe the total flux, i.e. they include the increases in both concentrations and
286
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
Fig. 1. The three sub-catchments – I, II and III – at the Balån study site.
runoff associated with clear-felling, thus showing the importance of the procedure stated above of separating the two effects in the calculations.
values of 87 and 72 kg ha−1 y−1 for periods 1 and 2, respectively [1.24 * 1.40*50= 87 kg ha− 1 y−1 and 1.15 * 1.25* 50= 72 kg ha−1 y−1]. 2.4. Forest management problem formulation and assumption
2.3.3. Nitrogen (N) and N-fertilization It is assumed that fertilization is performed at a standard dosage of 150 kg N ha−1, and that 5% of the nitrogen in applied fertilizer is emitted to streams, resulting in a 1%y−1 increase during the 5-year period following fertilizer application (Löfgren, 2007). Hence, an additional 7.5 kg N ha−1 is lost during the five years after N fertilization, corresponding to an additional annual loss of 1.5 kg Nha−1 y−1 [(5%/ 5 y)*150 kg Nha−1 =1.5 kg Nha−1 y−1)]. 2.3.4. Phosphorus (P) The estimated increase in P concentration is based on the data in Löfgren and Olsson (1990) and the conclusions in Löfgren (2007), which indicate that the P flux doubles during the three years following final felling compared with the base level of 0.06 kg ha− 1 y−1. This results in an estimated average flux of 0.096 kg P ha−1 y−1 during the first 5-year period [(0.06 * 2.0 * 3 + 0.06 * 2)/5 = 0.096 kg P ha−1 y−1]. 2.3.5. Methyl mercury (MeHg) The base level is set at 0.3 mg ha−1 y−1 following data from Balån presented by Sørensen et al. (2009; Table 2). It is assumed that final felling increases the total flux of MeHg 3-fold compared with an undisturbed, growing forest, and that this level persists for 10 years, which is consistent with the assumptions of Bishop et al. (2009). 2.3.6. Dissolved Organic Carbon (DOC) The base level for DOC, of 50 kg ha− 1 y−1, was set based on data from Balån presented by Laudon et al. (2009), and the increases resulting from final felling for the first and second 5-year periods are estimated, using the model presented by Öhman et al. (2009; Table 1, Case 1a), at 24% and 15% over the base level, respectively. Adding the effect of increased water flow, according to the model presented above, yields
The forest management problem consists of selecting a treatment schedule for every stand in the catchment so that the NPV is maximized subject to constraints imposed by maximum allowed substance concentrations, final forest stock and a defined maximum decrease in harvest volumes between periods. A treatment schedule is a sequence of forest operations – in this study thinning, fertilization, final felling and other silvicultural treatments – over the 100-year planning horizon for a particular stand. Thus, the management regulation method is entirely based on even-aged management. The NPV is based on the sum of revenues from and costs of timber harvest and costs of silviculture over the planning horizon, discounted with a 3% interest rate. The runoff concentration constraints are imposed for each 5-year period and each substance. Hence, increases in concentrations over the base level should be restricted to acceptable levels in each period for each substance (at sub-catchment level and the entire catchment, in the respective analyses). The increase in concentration of a substance in a specific 5-year period is, as explained above, a function of the area of final felling and N fertilization in that or the previous period. The average stocking should be at least 100 m3 ha−1 after 100 years in order to control the state of the forest at the end of the planning horizon. The harvest volume is allowed to decrease by a maximum of 10% per 5-year period; this constraint has negligible effect on the NPV, but results in solutions that are more realistic from a forestry perspective. Since we did not know which treatment schedules are optimal, we formulated a large set of schedules for each stand to choose from, which differ in the timing of thinning, final harvest, and fertilization. Optimal schedules, constituting the optimal solution, are identified by solving the mathematical model specified in the next section. Treatment schedules are simulated using the GAYA stand simulation system (Eriksson, 1983; Hoen and Eid, 1990). All permissible
Table 1 Forest data for the three sub-catchments I, II, and III and for the whole area. Sub-catchment
Area (ha)
Stand age (y)
Stem volume (m3 ha−1)
Norway spruce (%)
Scots pine (%)
Other (%)
Site index, H100 (m)
I II III I + II + III
37 45 156 238
90 97 41 61
163 158 74 104
50 38 41 42
40 53 36 40
9 9 24 19
17.4 16.9 18.6 18.1
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
287
Table 2 Annual unit area flux for undisturbed conditions (base level), at the first 5-year period after the forestry operation (1st period) and 6–10 years after the forestry operation (2nd period) and concentrations for N, P, MeHg and DOC under undisturbed conditions as well as from 100% clear-felled or fertilized area. Flux W (clear-felling) −2
Lm Base level 1st period 2nd period
y
−1
400 560 500
N (clear-felling) kg ha
−1
y
−1
1.5 3.0 3.0
N (N-fertilization) −1
kg ha
y
−1
P (clear-felling) −1
kg ha
y
−1
MeHg (clear-felling) −1
mg ha
y
−1
DOC (clear-felling) kg ha−1 y−1
0 1.5 0
0.060 0.096 0.060
0.3 0.9 0.9
50 87 72
mg L−1
mg L−1
mg L−1
ng L−1
mg L−1
0.375 0.536 0.600
– 0.750a –
15.0 17.1 12.0
0.075 0.161 0.180
12.5 15.5 14.4
Concentration
Base level 1st period 2nd period a
400 560 500
Base level water runoff is assumed.
schedules within the given specifications are generated for each stand, resulting in a total of 31,038 schedules, corresponding to an average of 470 schedules per stand. Standard treatments include stand establishment, precommercial thinning of young forest, thinning, fertilization, and final felling. The minimum age for final felling for each stand is set according to the Forestry Act of Sweden regarding corresponding site productivity (SNBF, 1994). Thinning treatments have an intensity of 30% removal of the basal area. Fertilization is at 150 kg of N ha−1. The GAYA simulator uses growth functions by Ekö (1985). Revenues are computed with functions from Ollas (1980), with timber prices according to the forest owners' organization Norra skogsägarna for the Umeå region, valid from August 2007. The quality distributions of pine and spruce logs are weighted with data from SOU (2007). Harvesting costs are computed based on functions presented by Nurminen et al. (2006) with the cost per hour set to 750 and 650 SEK for harvester and forwarder, respectively. The fertilization cost is set to 2300 SEK ha−1. Costs of other silvicultural activities are taken from SOU (2009). For the year 2007, the average amounted to 7600 SEK ha−1 for site preparation and planting and to 2300 SEK ha−1 for precommercial thinning. 2.5. Mathematical model The forest management problem described above was formulated as a linear programming (LP) model. An optimal solution to the problem, i.e. a solution that maximizes the NPV and satisfies the concentration, final stock, and harvest volume requirements, is a combination of treatment schedules (as outlined above) for each stand in the catchments. This combination determines the output values for N, P, MeHg and DOC over time for the entire area. The model formulation is known as a Model I formulation (Johnson and
Table 3 Scenario codes and assumptions. Scenarioa
Assumptions
Base
Concentrations of N, P, DOC and MeHg are allowed to increase ≤ 10%, respectively. Clear felling and fertilization are the only forestry operations assumed to affect the concentrations MeHg0.5 As scenario Base but the excess leakage of MeHg after harvest is reduced by 50% MeHgThinn As scenario Base – WsAll but the excess leakage of MeHg ha−1 is assumed to be the same after thinning as after final felling Unconstrained No constraints on concentration; otherwise as scenario Base a The scenario is applied either for each sub-catchment, denoted WsSep, or to the entire area, denoted WsAll.
Scheurman, 1977). In technical terms, a solution consists, for each stand i, i ∈ {1, …, I}, of the number of hectares that should be allocated to each treatment schedule j, j ∈ J(i), where J(i) is the set of treatment schedules pertaining to stand i. There are T periods and t ∈ {1, …, T}. Associated with each treatment schedule j for each stand i are the NPV ha−1, Rij, the harvesting volume ha−1 in period t, Hijt, the standing volume Vij ha−1 after harvest in the last period, an indicator Cijt set to 1 if the stand is clear felled in period t, otherwise 0, and an indicator Fijt set to 1 if the stand is fertilized in period t, otherwise 0. Each stand belongs to one of three sets, S(r), s ∈ {1, 2, 3} corresponding to the subcatchments. Given that the decision variable xij signifies the number of hectares of stand i managed according to treatment schedule j, the areas of final harvest and fertilization in period t in sub-catchment r (ctr and ftr, respectively) are given by ∑ ∑ Cijt ⋅xij = ctr
t = 1…; T; r = 1; …3
ð1Þ
∑ ∑ Fijt ⋅xij = ftr
t = 1…; T; r = 1; …3
ð2Þ
i∈Sðr Þ j∈J ðiÞ
i∈Sðr Þ j∈J ðiÞ
This means that the flux of water in period t from sub-catchment r, wtr, can be expressed as wtr = BWr + FW 0 −FW B ⋅ctr + FW 1 −FW B ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð3Þ
where BWr is the total flux of water from sub-catchment r under undisturbed conditions (base level), and FW0, FW1, and FWB are the flux of water ha−1 during the 5-year period when harvest is conducted, the flux one period after the harvest, and the base level flux, respectively (cf. Table 2). This means that the second term adds the additional flux above the base level following the harvest in period t and the third term gives the additional flux following the harvest in the previous period. Following the same logic, the flux of N in period t from sub-catchment r, ntr, can be given as: 0 B 1 B ntr = BNr + FN −FN ⋅ctr + FN −FN ⋅cðt−1Þr + FE ⋅ ftr t = 1…; T; r = 1; …3
ð4Þ
where BNr is the total flux of N from sub-catchment r under undisturbed conditions (base level), and FN0, FN1, and FNB are the flux of N ha−1 during the 5-year period when harvest is conducted, the flux one period after the harvest, and the base level flux, respectively. The fourth term in Eq. (4) is the additional flux stemming from the leakage of fertilizer, where FE is the flux of N from a fertilized ha. Subsequently, Eqs. (5)–(7) compute the total flux
288
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
in period t from sub-catchment r for P (ptr), MeHg (mtr), and DOC (dtr), respectively, through 0 B 1 B ptr = BPr + FP −FP ⋅ctr + FP −FP ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð5Þ
mtr = BMr + FM 0 −FM B ⋅ctr + FM1 −FMB ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð6Þ
dtr = BDr + FD0 −FDB ⋅ctr + FD1 −FDB ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð7Þ
The constraint on concentration is then formulated such that the flux of the substance divided by the water flux, i.e. the concentration, in a particular sub-catchment and period, should be less than or equal to the base level concentration multiplied by the allowed relative increase in concentration, δ. For N, P, MeHg, and DOC, respectively, this is given as ntr = wtr ≤ ð1+δÞ⋅BNtr = BWtr
t = 1…; T; r = 1; …3
ð8Þ
ptr = wtr ≤ ð1+δÞ⋅BPtr = BWtr
t = 1…; T; r = 1; …3
ð9Þ
mtr = wtr ≤ ð1+δÞ⋅BMtr = BWtr
t = 1…; T; r = 1; …3
ð10Þ
dtr = wtr ≤ ð1+δÞ⋅BDtr = BWtr
t = 1…; T; r = 1; …3
ð11Þ
(A slight transformation of Eqs. (8)–(11) will yield a linear relationship.) The given formulation refers to the requirements for each sub-catchment. When the concentration levels relate to the whole catchment area, the fluxes of substances are simply summed over the sub-catchments. The final stock demand is expressed by requiring the standing volume to be at least 100 m3 ha−1, i.e., I
∑ ∑ Vij xij ≥100⋅A
ð12Þ
i = 1 j∈J ðiÞ
where A is the total forest area. The harvest volume is not allowed to decrease by more than 10% from one period to the next, i.e., I
I
∑ ∑ Hijt xij ≥ ∑ ∑ 0:9⋅Hijðt−1Þ xij
i = 1 j∈J ðiÞ
i = 1 j∈J ðiÞ
t = 2…; T
ð13Þ
Treatment schedules must be allocated to the whole area of each stand, i.e., ∑ xij = Ai
i = 1…; I
j∈J ðiÞ
ð14Þ
period is set to 10%. This threshold was based on the assumption that concentrations are currently close to the limits between Good and Moderate status, which is true for P at Balån (Löfgren et al., 2009a), thus a 10% increase would result in a status shift from Good to Moderate. The scenario was applied either for each sub-catchment, denoted as case WsSep, or to the entire area, denoted as case WsAll. A sensitivity analysis was also conducted, in which the excess N and P fluxes in the period of final felling were doubled during the first 5-year period and zeroed during the second period (cf. Table 2). However, the analysis gave the same result as the base case and is not reported here. To test the influence of concentration levels on NPV, the model was solved for a range of values of δ, in addition to the base case scenario. The effects of imposing no concentration constraints were also tested, corresponding to solving the problem with a large value for δ. The scenarios are defined in Table 3. 3. Results 3.1. Economic implications As shown in Fig. 2, economic losses increase rapidly as the strictness of the concentration limitations increases. With restriction levels of 6% and 5% for WsSep and WsAll, respectively, less than half the economic value of the forest remains. An increase of ca. 30% in concentrations must be allowed if maximum NPV is to be retained. The loss for case WsSep is greater than for case WsAll (illustrating the scale effect), in accordance with expectations since in the latter case a high concentration of a substance in one sub-catchment can be compensated by a low concentration in another sub-catchment. From Table 4 it is evident that the assumptions regarding MeHg have a very large impact on the economic result. If the flux of MeHg is reduced by 50% compared to the Base scenario, the NPV value increases by 21% and 41% for cases WsAll and WsSep, respectively. The assumption that thinning would have the same effect as final felling appears to be less important regarding MeHg. 3.2. Concentrations In the Base and MeHgThinn scenarios, MeHg is always the limiting substance; no other substance reaches the 10% excess level in these cases (Table 5). In the MeHg0.5 scenario, however, N becomes the limiting element. Both MeHg and N could reach values well above the 10% level in sub-catchments in case WsAll, since the constraint is valid only for the area as a whole. P is always at a low level, whereas in some periods DOC could reach or exceed the 10% level for some subcatchments in case WsAll. When MeHg is limiting, it is limiting for all sub-catchments and all 5-year periods, except the first two, in case WsSep. In case WsAll it is limiting for the whole area for most periods. WsAll
where Ai is the area of stand i and the x-variables must be nonnegative, i.e.,
WsSep
100% 80%
xij ≥ 0
i = 1…; I; j = 1…; J ðiÞ
ð16Þ 60%
Finally, the NPV is maximized, i.e. 40% I
Max Z1 = ∑ ∑ Rij xij i = 1 j∈J ðiÞ
ð15Þ
2.6. Scenarios
20% 0% 0% -20%
The base case scenario represents the NPV when the model is solved with a value for δ of 0.1, i.e. the maximum allowed increase in concentration for any substance above the reference level in any
10%
20%
30%
40%
50%
60%
δ
Fig. 2. The NPV relative to its maximum value for different maximum allowed relative concentration levels (δ) for cases WsAll and WsSep.
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
(a)
Table 4 The relative NPV of the different scenarios compared with scenario Base (= 100; g%).
289
I+II+III
I
II
III
0.120
WsAll MeHg0.5 MeHgThinn
121 88
0.115 0.110
WsSep MeHg0.5 MeHgThinn
0.105
141 90
0.100 0.095 0.090
The only period it is below 9% is the first period. When N is limiting (scenario MeHg0.5) the N level is at 10% in at least one sub-catchment for the 2nd to the 12th periods in case WsSep, and at the 10% level for the whole area in case WsAll for the 2nd to the 8th period (Fig. 3).
0.085 0.080 0.075 0.070
3.3. Forest management
(b)
P WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstrained MeHg WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstrained DOC WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstrained
I
II
7
9
11
13
15
17
19
I+II+III
I
II
III
0.51
Table 5 The highest concentration of a substance above the base level (%) found in any period measured in the sub-catchments separately (I, II and III) or in the area as a whole (I–III) for the different scenarios and the cases WsAll and WsSep. I–III
5
5-year period
The comparatively low site index of the area is reflected in the forestry activities of the solution to the Unconstrained scenario. Stands are finally felled after a rotation period of ca. 100 years and
N WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstraineda
3
1
0.49 0.47
III
0.45 7.0 10.0 6.5
4.2 10.0 2.0
5.6 10.0 2.0
8.5 10.0 1.9
5.5 10.0 4.9 16.1
21.1 33.7 23.1 35.4
18.4 29.9 19.7 33.3
6.1 14.7 6.1 16.3
0.43 0.41 0.39 0.37 0.35 1
3
5
7
9
11
13
15
17
19
5-year period 0.3 0.7 0.3
0.8 1.9 2.0
1.0 1.5 2.0
1.0 1.5 1.9
0.7 1.1 0.6 1.1
3.2 5.5 3.6 5.5
2.8 5.0 2.4 4.5
0.8 1.2 0.6 1.5
10.0 2.7 10.0
10.0 3.3 10.0
10.0 2.8 10.0
10.0 2.6 10.0
10.0 2.7 10.0 25.4
52.2 8.7 57.2 64.4
45.9 8.2 48.8 68.2
15.1 3.1 15.1 27.6
Fig. 3. The concentrations for each sub-catchment I–III and the whole area with case WsAll: (a) MeHg (ng L−1) under scenario Base and (b) N (mg L−1) under scenario MeHg0.5.
they are not thinned more than once per tree generation (cf. 2.4 ha of final felling with 1.7 ha of thinning y−1; see Table 6). The final felling activity drops as a result of the 10% concentration constraints in the Base scenario; a change that is somewhat compensated for by increased thinning. Fertilization also decreases, but the operation is not limited by the N concentration threshold. Instead, it is coupled to the reduced harvest activity. In the MeHg0.5 scenario, final felling increases to the levels of the Unconstrained scenario. Fertilization is reduced (WsAll) or not affected (WsSep) compared with the Base scenarios, since N now becomes a limiting constraint. In the Table 6 The average ha−1 y−1 of different forestry operations over the 100 years planning horizon for the different scenarios and the cases WsSep and WsAll.
1.7 4.0 1.6
2.0 5.0 2.0
2.0 4.3 2.0
2.0 3.9 1.9
1.7 3.9 1.6 4.0
8.2 11.7 9.0 12.2
7.5 10.3 7.7 10.6
2.5 4.6 2.4 4.6
a Note that the solution to Unconstrained is equal for WsSep and WsAll and therefore only reported for one of the cases.
Case
WsSep
WsAll
Final Fertilization Thinning Final Fertilization Thinning felling felling BaseScenario MeHg0.5 MeHgThinn Unconstraineda a
1.5 2.4 1.5 2.4
0.8 0.8 0.9 1.3
1.9 1.8 0.0 1.7
1.4 2.2 1.4 2.4
Note that Unconstrained is equal for WsSep and WsAll.
0.5 0.8 0.3 1.3
1.8 1.4 0.0 1.7
290
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
Ha
(a) 10 9 8 7 6 5 4 3 2 1 0
I
II
III
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
5-year period
Ha
(b) 10 9 8 7 6 5 4 3 2 1 0
I
II
III
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
5-year period Fig. 4. The final felling area in the sub-catchment areas I, II and III area over time in scenario Base for (a) WsSep and (b) WsAll.
MeHgThinn scenario, thinning activities cease due to the elevated MeHg concentrations. The reduced area of final felling in the Base compared with the Unconstrained scenario roughly agrees with the drop in economic value (NPV) for case WsSep; there are ca. 65% declines in both variables. The drop in economic value is smaller in case WsAll, because the final felling can be allocated over time in a more profitable manner in WsAll than in WsSep, as illustrated by the markedly different patterns of harvest area allocation among sub-catchments in Fig. 4. The harvest patterns for WsSep and WsAll of the BaseScenario show that a greater volume can be extracted earlier in case WsAll than in WsSep, leading to a higher economic return (Fig. 5). 4. Discussion The results indicate that the requirements to maintain or improve water quality stipulated by the EU Water Framework Directive goals may influence forest management, both in how the forest is managed and the economic consequences. The results agree on a general level with Öhman et al. (2009), in that maintenance of water quality may
m3
WsSep
WsAll
5000 4500 4000 3500 3000 2500 2000 1500 1000 500 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
5-year period Fig. 5. The harvest volume over time for cases WsAll and WsSep of scenario Base.
have an important effect on the economic outcome of forestry if it is implemented as a constraining factor. According to the present study, the consequences for forest management would be changes to the timing and reductions in the extent of final felling. Additionally, N fertilization would be reduced, even though this activity is not constrained by the N concentration limitation but a consequence of the reduced area of final felling, since N fertilization is most profitable when it precedes harvest. The results also show that there is a spatial scale effect (see e.g. Fig. 3 and Table 5). This is to be expected for drainage basins like this, where the age structure of stands differs substantially amongst the sub-catchments and there is much to gain from allocating the forest operations differently over time. However, if criteria for water quality are implemented for small sub-catchments or even-aged large catchments, this will not be an uncommon situation. The results thus indicate that it is important to identify critical points in the stream system where the WFD concentration constraints may be exceeded and then formulate restrictions on concentrations for those target points, a procedure suggested by Öhman et al. (2009). The results are highly dependent on the assumptions regarding responses of MeHg fluxes to forest management, for two main reasons: clear-felling increases MeHg fluxes 3-fold, but N fluxes only 2-fold and the effect persists for two 5-year periods. This explains not only why MeHg is generally the limiting substance, but also its cyclical pattern with clear-felling. When MeHg flux is also assumed to be caused by thinning this treatment essentially disappears. It should be noted in this context that thinning is not assumed to increase water runoff, only the concentration and, hence, the MeHg flux, thus it contributes even more strongly than final felling to the concentration. The economic implications of discontinuing thinning are limited, because of the fairly low fertility of the Balån area. MeHg concentrations (range; 0.077–0.126 ng MeHg L−1 for all scenarios and sites) under the given assumptions are 2- to 6-fold lower than those measured in Balån (Sørensen et al., 2009). Since the constraints are formulated as relative concentrations, this discrepancy does not affect the general conclusions, but it could be important in practice if absolute concentrations are considered and threshold values implemented. The concentration levels are low compared with those found in 109 randomly selected headwater forest streams in the River Dalälven catchment in central Sweden by Löfgren (25- and 75percentiles; 0.16 and 0.49 ng L−1, respectively, unpublished data). For all those reasons sensitivity analyses testing the assumptions concerning MeHg would be valuable. Furthermore, some of the data that the assumptions regarding forestry effects on N and P are based upon were collected in the 1970s and 1980s (Löfgren et al., 2009b). Since then water quality has been identified as a vital issue for the forestry sector and recommendations to improve forestry operations in order to protect surface water quality have been issued. Thus, on average, the fluxes of these two substances may be overestimated. However, results of observations during the first two years after clear-felling with modern logging techniques at Balån (Löfgren et al., 2009b) indicate that the current model assumptions are relevant with regard to the short-term N and P losses for the first 5-year period. In the management model, it is implicitly assumed that all forest operations take place at the beginning of the 5-year period. If instead they were assumed to take place evenly over the entire period, the effects would be distributed over a slightly longer period, which in turn may make it slightly easier to avoid concentration peaks. Still, it is unlikely that this would have any great influence on the results. Furthermore, in practice forestry operations tend to concentrate harvest activities as this reduces costs of hauling machinery and road maintenance. In this study all considered substances have been assumed equally important and the same thresholds, in percentage increases have been set for all of them. An alternative approach would have been to
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
judge the relative severity of excess concentrations of each substance, expressed as value functions, and weigh them accordingly. This would mean that the problem would be dealt with as a multiple criteria problem. This would be reasonable, as there are so many uncertainties and subjective judgments that have to be made in cases where the scientific data are insufficient. These uncertainties are reality for the water authorities, who have to take decisions on the ecological status in each water body, regardless of data quality. At present, the ecological status of 44% of the Swedish stream water bodies (n = 15563) is at risk of not achieving Good ecological status (Water Authorities, 2010). In many of those streams, forestry is the only anthropogenic impact. The selected concentration restrictions in this study are not connected to the Swedish surface water classification system (SEPA, 2007), which besides biological and hydromorphological parameters includes WFD constraints for only acidity (pH) and phosphorus. However, the classification system is being continuously developed, and there are demands from society to include more chemical parameters (such as N, DOC and MeHg) and to improve monitoring (Water Authorities, 2010). Hence, the selected concentration restrictions in this study are used as examples of the principle of using multiple indicator constraints as supplementary input variables for forest management. With the exception of DOC, the assumed concentration restriction of an additional 10% is low compared to the normal variation of these substances in pristine forests. Based on annual means of concentrations measured from 1996 to 2008 at three forested, non-managed integrated monitoring sites (Aneboda, Kindla and Gammtratten) the coefficient of variation (CV) varies between 16 and 35%, 31–38%, 9–12% and 17–37% for nitrogen, phosphorus, DOC and Hg (note not MeHg), respectively (Löfgren unpublished data). This large inter-annual variation is rarely taken into account by the water authorities due to lack of data, and the classifications are mostly based on a few single observations or short-term time series (Water Authorities, 2010). Hence, there are large uncertainties in the classifications, which do not take into account the temporal variations, and the risks of misclassifications are obvious, especially close to the limits between High and Good status or Good and Moderate status (Löfgren et al., 2009a). Exceeding any of these limits due to a 10% concentration increase could induce remedial actions from the water authorities. Hence, it seems unlikely that the water authorities would use the simulated concentration increases to impose restrictions on forest management in the studied sub-catchments unless the water quality was very close to the class limit between Good and Moderate status, which only applies to P (Löfgren et al., 2009a). The broad conclusion from this study is that the WFD may indeed affect forestry if the effects of forest operations on water quality are sufficiently strong. The more quantitative results must, however, be interpreted with the utmost caution. This is merely a first attempt at analyzing the diverse dimensions of water quality that are affected by forest management, and great uncertainties still remain regarding the cause and effect relationships. We also need to consider dimensions other than those addressed here, such as acidification, erosion of particles and biological effects. However, water quality protection will ultimately depend largely on how well the forest operations are planned and carried out in the field, and a model such as the one described here can help in the planning process. Acknowledgements We would like to thank Jakob Schelker for supplying data from the Balån catchment and Anu Hankala, both at SLU, for preparation of the maps. The research has been supported by the Kempe Foundations and the Swedish University of Agricultural sciences, SLU.
291
References Bishop, K., Allan, C., Bringmark, L., Garcia, E., Hellsten, S., Högbom, L., Johansson, K., Lomander, A., Meili, M., Munthe, J., Nilsson, M., Porvari, P., Skyllberg, U., Sørensen, R., Zetterberg, T., Åkerblom, S., 2009. The effects of forestry on Hg bioaccumulation in nemoral/boreal waters and recommendations for good silvicultural practice. Ambio 38 (7), 373–380. Ekö, P.M., 1985. A growth simulator for Swedish forests, based on data from the national forest survey. Report No. 16. Department of Silviculture, SLU. Umeå. (In Swedish). Eriksson, L.O., 1983. Timber class formation by cluster analysis. Swedish University of Agricultural Sciences, Department of Forest Technology. Report No. 157. Garpenberg Sweden. Hagner, O., 1990. Computer aided stand delineation and inventory based on satellite remote sensing. In: Sylvander, R., Poso, S. (Eds.), The Usability of Remote Sensing for Forest Inventory and Planning. : Report No. 4. Remote Sensing Laboratory, SLU. Umeå. Hoen, H.F., Eid, T., 1990. En modell for analyse av behandlingsstrategier for en skog ved bestandssimulering og lineær programmering (A model for analysis of treatment strategies for a forest by stand simulation and linear programming). Rapport/Norsk institutt for skogforskning – NISK, 90:9, Ås. . (In Norwegian). Johnson, K.N., Scheurman, H.L., 1977. Techniques for Prescribing Optimal Timber Harvest and Investment Under Different Objectives – Discussion and Synthesis: Forest Science, Monograph, 18 (Supplement to Number 1, 1 March 1977), pp. a0001-z0001(1). Kreutzweiser, D.P., Hazlett, P.W., Gunn, J.M., 2008. Logging impacts on the biogeochemistry of boreal forest soils and nutrient export to aquatic systems: a review. Environ. Rev. 16, 157–179. Laudon, H., Hedtjärn, J., Schelker, J., Bishop, K., Sørensen, R., Ågren, A., 2009. Response of dissolved organic carbon following forest harvesting in a boreal forest. Ambio 38 (7), 381–386. Löfgren, S., 2007. Conclusions from the workshop. In: Löfgren, S. (Ed.), How to estimate N and P losses from forestry in northern Sweden: KSLAT 146:24–30. ISSN 0023– 5350. Löfgren, S., Olsson, H., 1990. Input of nitrogen and phosphorus to running waters in the Swedish inland. Background document to Marine'90, Action plan against marine eutrophication. Swedish Environmental Protection Agency report 3692. 100 pp., (In Swedish). Löfgren, S., Kahlert, M., Johansson, M., Bergengren, J., 2009a. Classification of two Swedish forest streams in accordance with the EU Water Framework Directive. Ambio 38 (7), 394–400. Löfgren, S., Ring, E., von Brömssen, C., Sørensen, R., Högbom, L., 2009b. Short-term effects of clear-cutting on the water chemistry of two boreal streams in northern Sweden: a paired catchment study. Ambio 38 (7), 347–356. Nisbet, T., 2001. The role of forest management in controlling diffuse pollution in UK forestry. For. Ecol. Manage. 143, 215–226. Nurminen, T., Korpunen, H., Utrotningsitalo, J., 2006. Time consumption analysis of the mechanized cut-to-length harvesting system. Silva Fenn. 40 (2). Öhman, K., Seibert, J., Laudon, H., 2009. An approach for including consideration of stream water Dissolved Organic Carbon in long term forest planning. Ambio 38 (7), 387–394. Ollas, R., 1980. New yield functions for trees and stands. Forskningsstiftelsen Skogsarbeten, Ekonomi no 5. 4 pp. Stockholm. . (In Swedish). Piirainen, S., Finér, L., Mannerkoski, H., Starr, M., 2007. Carbon, nitrogen and phosphorus leaching after site preparation at a boreal forest clear-cut area. For. Ecol. Manage. 243, 10–18. Reese, H., Nilsson, M., et al., 2003. Countrywide estimates of forest variables using satellite data and field data from the National Forest Inventory. Ambio 32 (8), 542–548. Ring, E., Löfgren, S., Sandin, L., Högbom, L., Goedkoop, W., 2008. Forestry and water – a review. Skogforsk Redogörelse Nr 3 2008. . (in Swedish with English summary). SEPA, 2007. Status, potential and quality criteria for lakes, streams, coastal waters and estuaries – a handbook on how the quality criteria in water bodies can be determined and monitored. Swedish Environmental Protection Agency report 2007:4. 414 pp., (In Swedish). SEPA, 2008. Statute book of the Swedish Environmental Protection Agency NFS 2008:1 (regulations and recommendations on the classification of and environmental quality standards for surface waters). ISBN 1403–8234, 116 pp. (In Swedish). SNBF, 1994. Swedish forestry act. Swedish National Board of Forestry (SNBF), Jönköping, Sweden. Sørensen, R., Meili, M., Lambertsson, L., von Brömssen, C., Bishop, K., 2009. The effects of forest harvest operations on mercury and methylmercury in two boreal streams: relatively small changes in the first two years prior to site preparation. Ambio 38 (7), 364–372. SOU, 2007. Forest statistics – Annual report 2007. Swedish Forest Agency, Jönköping. SOU, 2009. Forest statistics – Annual report 2009. Swedish Forest Agency, Jönköping. http://www.svo.se/episerver4/default.aspx?id=37802. Swedish Forest Agency, 2000. Guidelines and foundations for the Forest Agency work on issues regarding forestry and water. Swedish Forest Agency Cirkulär 2000:5. 7 pp., (In Swedish). Swedish Forest Stewardship Council, 1998. http://www.fsc-sweden.org/Portals/0/Fsc-eng. pdf_1.pdf 1998. Water Authorities, 2010. In: Egerup, J., Eriksson, A., Göranson, M., Tornevall, H. (Eds.), Water – Our Common Responsibility: Vattenmyndigheterna, Tabergs tryckeri, March 2010, p. 30.
Contents lists available at ScienceDirect
Forest Policy and Economics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / f o r p o l
Implications for forest management of the EU Water Framework Directive's stream water quality requirements — A modeling approach Ljusk Ola Eriksson a,⁎, Stefan Löfgren b, Karin Öhman a a b
Swedish University of Agricultural Sciences, Department of Forest Resource Management, S-901 83 Umeå, Sweden Swedish University of Agricultural Sciences, Department of Aquatic Sciences and Assessment, P.O. Box 7050, S-750 07, Uppsala, Sweden
a r t i c l e
i n f o
Article history: Received 17 June 2010 Received in revised form 24 December 2010 Accepted 3 February 2011 Available online 15 March 2011 Keywords: Long-term planning Linear programming EU Water Framework Directive Nutrients Methyl mercury Dissolved organic carbon
a b s t r a c t The EU Water Framework Directive (WFD) stipulates that measures should be taken to ensure that all lakes and streams in the EU have good ecological and chemical status, comparable to that of waters unaffected by human activities. This has profound potential implications for forestry, since operations such as harvesting and fertilization tend to reduce the quality of ground and stream water in affected catchments. The aim of this study is to assess the implications for forestry of limiting the concentrations of chemical substances reaching lakes and streams. A forest planning model with a horizon of 100 years that includes requirements regarding water concentrations of nitrogen (N), phosphorus (P), methyl mercury (MeHg) and dissolved organic carbon (DOC) was applied to three intensively-studied sub-catchments at Balån in the boreal part of northern Sweden. Limiting maximum increases in concentrations of these substances to 10% above reference values resulted in an economic loss of ca. 20% or 35%, depending on whether the limits were applied to the whole area or to each sub-catchment individually. The results were also highly dependent on the assumptions, especially regarding the flux of MeHg. The results should be interpreted with caution as there are still major uncertainties concerning the cause and effect relationships. We also need to consider additional aspects to those addressed here, such as acidification, erosion and the biological effects of the operations. © 2011 Elsevier B.V. All rights reserved.
1. Introduction The EU Water Framework Directive (WFD, 2000/60/EC) has set an objective to ensure that by 2015 the ecological and chemical status of all lakes and streams in Europe should be ‘good’, defined in terms of comparability to the water quality in pristine areas, unaffected by human activities. This is likely to have consequences for forestry operations at both stand and catchment levels, since hydrological, biological and chemical processes are all influenced by forestry activities (in addition to forest yields). Most importantly, in this context, operations such as forest harvesting and fertilization tend to reduce the water quality of streams and lakes receiving water from the forests, relative to those receiving water from non-managed forests (Piirainen et al., 2007; Kreutzweiser et al., 2008). However, the magnitude of these effects is clearly related to the proportion of the catchment affected by forestry operations and how these operations are performed (Ring et al., 2008). In Sweden, implementation of the WFD is led by five water authorities, which are authorized to apply management plans to enhance water quality, including plans to address effects of forestry, ⁎ Corresponding author at: Department of forest resource management, SLU, S-901 83 UMEÅ, Sweden. Tel.: + 46 706440004; fax: + 46 90778116. E-mail addresses: [email protected] (L.O. Eriksson), [email protected] (S. Löfgren), [email protected] (K. Öhman). 1389-9341/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.forpol.2011.02.002
where necessary to meet WFD requirements. A key task for these authorities is to assess the ecological status of waters. For this purpose, the Swedish Environmental Protection Agency (SEPA) has developed a classification system, which includes biological, chemical and hydro-morphological variables (SEPA, 2007, 2008). Hence, knowledge of the effects of forestry activities on values of these parameters is essential for understanding the adaptations forestry will need to make to meet the stipulations of the WFD as interpreted by the Swedish water authorities. Based on the SEPA system, the water authorities classify the ecological status as less than good in 39% and 44% of Swedish lakes and streams, respectively (Water Authorities, 2010). One reason for this is that the chemical status of the water in many forest streams tends to be close to the border between good and moderate classes (Löfgren et al., 2009a), which is the threshold for remedial actions according to the WFD. In many of these limnic systems, forestry is the major human influence, and a small increase in (for instance) phosphorus concentration caused by forestry might induce a shift from good to moderate class, and thus initiate a requirement for remedial actions. Another contributory factor, in some cases, may be incorrect classification based on the biological indices (Löfgren et al. op. cit., see below). Guidelines for mitigating the adverse effects of forest operations on water quality have been issued by the Swedish Forest Agency (2000) and the Swedish Forest Stewardship Council (1998). Ways in which forestry could adapt to meet water quality requirements have
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
not been thoroughly investigated either. Some “rules of thumb”, such as not allowing more than 30% of a catchment to be harvested (Ring et al., 2008), have been proposed, but it is not clear if they would be either optimal or sufficient. Further, one of few relevant studies under boreal conditions (Öhman et al., 2009) indicated that ensuring that the concentration of dissolved organic carbon (DOC) in streams and lakes fed by groundwater to b10% above the reference value could incur economic losses (relative to current practices) of 5–30%, depending on the assumed magnitude of forestry effects on water quality variables. A more comprehensive understanding of the relation between water quality and forestry than Öhman et al. (2009) is attempted here by allowing several modeled water quality variables to be simultaneously affected by forest operations. However, limitations in knowledge of the relationships between forestry activities and water quality parameters preclude analysis of many variables. Biological effects are inherently difficult to assess, not the least because the lack of casual relationship between organism status and water quality. For example, it is not possible to distinguish between natural acidity and acidification caused by human activity from the status of benthic fauna and fish (Löfgren et al., 2009a). Hydromorphological disturbances in the study area are basically of historical origin and less related to present forestry (Löfgren et al., 2009a, see below). Hence, the water quality variables that will be analyzed in this report are limited to water chemistry. In the study area, it is possible to estimate initial chemical parameters of the streams, and to make reasonable quantified assessments of the responses of these parameters to forest activities. The focus is on forestry as a large scale, long-term activity, i.e. the perspective is strategic. Hence, some effects of forestry activities on water quality, such as machinery crossing streams during harvesting operations, are not included as they are dependent on specific operational plans. These aspects, important as they may be (Nisbet, 2001; Kreutzweiser et al., 2008; Ring et al., 2008), are assumed to be accounted for in consistency with the values assigned by the water quality models used in this study. The strategic perspective also reduces the importance of hydro-morphological parameters as they are related more strongly to measures intended to improve the water quality (restoration of rafting channels for example), than normal forest operations. However it should be noted that there are exceptions, like for instance the abundance of large woody debris in streams caused by normal forest operations. The aim of this study was to assess, in a strategic setting, the potential implications for forestry of limiting the concentrations of selected chemical substances in surface waters below certain threshold values over time. To do this we used a traditional forest planning model aimed at maximizing the net present value (NPV) from future harvest activities subject to traditional forest constraints and requirements regarding water quality. The water quality variables are concentrations of nitrogen (N), phosphorus (P), methyl mercury (MeHg) and dissolved organic carbon (DOC). Model outputs pertain to financial values, forest management activities and concentrations of the substances mentioned above during the next 100 years. The analysis focuses on three intensively studied sub-catchment areas at Balån in the boreal part of northern Sweden (Löfgren et al., 2009b). Further, since water quality (like most ecological phenomena) is dependent on spatial scale, the relationships between forestry activities and water quality are assessed at two geographical scales: sub-catchment and whole catchment.
285
20°15′ E, Fig. 1). The area was chosen since it has been surveyed since 2004 and much of the data needed were available from the area (see articles in the special issue on Balån in Ambio Vol. 38, No. 7, Nov. 2009). The area consists of a total of 66 stands distributed on three sub-catchments I, II and III (Fig. 1). Areas I and II contain stands that on average are considerably older than those in area III, whereas the three stands together have a fairly even age distribution. The catchment is dominated by Scots pine (Pinus sylvestris) and Norway spruce (Picea abies) with limited occurrence of other species, mostly birch (Betula spp.) (Table 1). The forest data for the area were estimated by the k nearest neighbor method (kNN) (Reese et al., 2003), with stands delineated with the algorithm developed by Hagner (1990). Note that the description of the area given by Löfgren et al. (2009b) is based on forest data obtained from the land owner and a stump survey, not completely in agree with the kNN data. This is reflected, for instance, in the site index, which is lower on average in the kNN data. 2.2. Water quality indicators Changes in concentrations of relevant substances can be used as indicators of the effects of specific forest operations, and can be calculated by dividing their respective fluxes by the water runoff before and after the operations. For example, if the N flux at base level is 1.5 kg ha−1 y−1 and the annual water runoff is 400 mm y−1 (corresponding to an annual runoff volume of 0.4 m3 water from each square meter of the catchment), the N concentration in the runoff water will be 0.375 mg N L−1 [(1.5 * 106 mg N/104 m2)/ 0.4 m = 375 mg N m−3 = 0.375 mg N L−1]. Additionally, if 10% of the catchment area is clear-felled, leaching 3.0 kg N ha−1 y−1 and runoff 560 mm water y−1, the concentration in water from the clear-felled area will be (3.0 * 106 mg N/104 m2)/0.56 m = 0.536 mg N L−1] and from the entire catchment 0.391 mg N L−1 [=0.375 * 0.9 + 0.536 * 0.1]. This procedure has been followed here, since in most scientific literature the excess loss of a substance coupled to a certain forest operation is usually expressed as annual flux per hectare. Since the functions for growth and yields used in the study operate with a time step of 5 years, the description of the effects on water quality is also given as averages for 5-year periods. Further, in the study it is assumed that the consequences of a forest activity last for at most two 5-year periods, i.e. 10 years. 2.3. Establishing model parameters The model parameters used in this study are presented in Table 2, which summarizes the average effect per year of various forest operations on selected substances, as well as the base levels in non-managed forests. The rationale for setting these parameters is given below. The increased flux values for all operations except fertilization are based on the rate of estimated increase after a 1 ha harvest (clear-felling). In the case of fertilization, the effect is from fertilizing 1 ha. 2.3.1. Water runoff (W) The base level runoff is set to 400 mm y−1, following data from Balån presented by Löfgren et al. (2009b; Table 3). The increase from final felling is estimated with the model presented by Öhman et al. (2009; Table 1, Case 1a), resulting in an increase in the first 5-year period of 40% and in the second 5-year period of 25% over the base level. Increased water flow following harvest has been assumed when calculating the post-harvest (1st and 2nd periods) concentrations for all substances.
2. Materials and methods 2.1. Study area We set parameter values for our model based on data derived from the Balån study area of north-eastern Sweden (Löfgren et al., 2009a, b), situated ca. 60 km from the coast in the county of Västerbotten (63°49′ N,
2.3.2. Nitrogen (N) and clear-felling The data given by Löfgren and Olsson (1990) and the conclusions in Löfgren (2007) indicate that clear-felling increases N fluxes from about 1.5 to 3.0 kg N ha−1 y−1 on average over an entire period of 10 years across the Bothnian Bay region. These values describe the total flux, i.e. they include the increases in both concentrations and
286
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
Fig. 1. The three sub-catchments – I, II and III – at the Balån study site.
runoff associated with clear-felling, thus showing the importance of the procedure stated above of separating the two effects in the calculations.
values of 87 and 72 kg ha−1 y−1 for periods 1 and 2, respectively [1.24 * 1.40*50= 87 kg ha− 1 y−1 and 1.15 * 1.25* 50= 72 kg ha−1 y−1]. 2.4. Forest management problem formulation and assumption
2.3.3. Nitrogen (N) and N-fertilization It is assumed that fertilization is performed at a standard dosage of 150 kg N ha−1, and that 5% of the nitrogen in applied fertilizer is emitted to streams, resulting in a 1%y−1 increase during the 5-year period following fertilizer application (Löfgren, 2007). Hence, an additional 7.5 kg N ha−1 is lost during the five years after N fertilization, corresponding to an additional annual loss of 1.5 kg Nha−1 y−1 [(5%/ 5 y)*150 kg Nha−1 =1.5 kg Nha−1 y−1)]. 2.3.4. Phosphorus (P) The estimated increase in P concentration is based on the data in Löfgren and Olsson (1990) and the conclusions in Löfgren (2007), which indicate that the P flux doubles during the three years following final felling compared with the base level of 0.06 kg ha− 1 y−1. This results in an estimated average flux of 0.096 kg P ha−1 y−1 during the first 5-year period [(0.06 * 2.0 * 3 + 0.06 * 2)/5 = 0.096 kg P ha−1 y−1]. 2.3.5. Methyl mercury (MeHg) The base level is set at 0.3 mg ha−1 y−1 following data from Balån presented by Sørensen et al. (2009; Table 2). It is assumed that final felling increases the total flux of MeHg 3-fold compared with an undisturbed, growing forest, and that this level persists for 10 years, which is consistent with the assumptions of Bishop et al. (2009). 2.3.6. Dissolved Organic Carbon (DOC) The base level for DOC, of 50 kg ha− 1 y−1, was set based on data from Balån presented by Laudon et al. (2009), and the increases resulting from final felling for the first and second 5-year periods are estimated, using the model presented by Öhman et al. (2009; Table 1, Case 1a), at 24% and 15% over the base level, respectively. Adding the effect of increased water flow, according to the model presented above, yields
The forest management problem consists of selecting a treatment schedule for every stand in the catchment so that the NPV is maximized subject to constraints imposed by maximum allowed substance concentrations, final forest stock and a defined maximum decrease in harvest volumes between periods. A treatment schedule is a sequence of forest operations – in this study thinning, fertilization, final felling and other silvicultural treatments – over the 100-year planning horizon for a particular stand. Thus, the management regulation method is entirely based on even-aged management. The NPV is based on the sum of revenues from and costs of timber harvest and costs of silviculture over the planning horizon, discounted with a 3% interest rate. The runoff concentration constraints are imposed for each 5-year period and each substance. Hence, increases in concentrations over the base level should be restricted to acceptable levels in each period for each substance (at sub-catchment level and the entire catchment, in the respective analyses). The increase in concentration of a substance in a specific 5-year period is, as explained above, a function of the area of final felling and N fertilization in that or the previous period. The average stocking should be at least 100 m3 ha−1 after 100 years in order to control the state of the forest at the end of the planning horizon. The harvest volume is allowed to decrease by a maximum of 10% per 5-year period; this constraint has negligible effect on the NPV, but results in solutions that are more realistic from a forestry perspective. Since we did not know which treatment schedules are optimal, we formulated a large set of schedules for each stand to choose from, which differ in the timing of thinning, final harvest, and fertilization. Optimal schedules, constituting the optimal solution, are identified by solving the mathematical model specified in the next section. Treatment schedules are simulated using the GAYA stand simulation system (Eriksson, 1983; Hoen and Eid, 1990). All permissible
Table 1 Forest data for the three sub-catchments I, II, and III and for the whole area. Sub-catchment
Area (ha)
Stand age (y)
Stem volume (m3 ha−1)
Norway spruce (%)
Scots pine (%)
Other (%)
Site index, H100 (m)
I II III I + II + III
37 45 156 238
90 97 41 61
163 158 74 104
50 38 41 42
40 53 36 40
9 9 24 19
17.4 16.9 18.6 18.1
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
287
Table 2 Annual unit area flux for undisturbed conditions (base level), at the first 5-year period after the forestry operation (1st period) and 6–10 years after the forestry operation (2nd period) and concentrations for N, P, MeHg and DOC under undisturbed conditions as well as from 100% clear-felled or fertilized area. Flux W (clear-felling) −2
Lm Base level 1st period 2nd period
y
−1
400 560 500
N (clear-felling) kg ha
−1
y
−1
1.5 3.0 3.0
N (N-fertilization) −1
kg ha
y
−1
P (clear-felling) −1
kg ha
y
−1
MeHg (clear-felling) −1
mg ha
y
−1
DOC (clear-felling) kg ha−1 y−1
0 1.5 0
0.060 0.096 0.060
0.3 0.9 0.9
50 87 72
mg L−1
mg L−1
mg L−1
ng L−1
mg L−1
0.375 0.536 0.600
– 0.750a –
15.0 17.1 12.0
0.075 0.161 0.180
12.5 15.5 14.4
Concentration
Base level 1st period 2nd period a
400 560 500
Base level water runoff is assumed.
schedules within the given specifications are generated for each stand, resulting in a total of 31,038 schedules, corresponding to an average of 470 schedules per stand. Standard treatments include stand establishment, precommercial thinning of young forest, thinning, fertilization, and final felling. The minimum age for final felling for each stand is set according to the Forestry Act of Sweden regarding corresponding site productivity (SNBF, 1994). Thinning treatments have an intensity of 30% removal of the basal area. Fertilization is at 150 kg of N ha−1. The GAYA simulator uses growth functions by Ekö (1985). Revenues are computed with functions from Ollas (1980), with timber prices according to the forest owners' organization Norra skogsägarna for the Umeå region, valid from August 2007. The quality distributions of pine and spruce logs are weighted with data from SOU (2007). Harvesting costs are computed based on functions presented by Nurminen et al. (2006) with the cost per hour set to 750 and 650 SEK for harvester and forwarder, respectively. The fertilization cost is set to 2300 SEK ha−1. Costs of other silvicultural activities are taken from SOU (2009). For the year 2007, the average amounted to 7600 SEK ha−1 for site preparation and planting and to 2300 SEK ha−1 for precommercial thinning. 2.5. Mathematical model The forest management problem described above was formulated as a linear programming (LP) model. An optimal solution to the problem, i.e. a solution that maximizes the NPV and satisfies the concentration, final stock, and harvest volume requirements, is a combination of treatment schedules (as outlined above) for each stand in the catchments. This combination determines the output values for N, P, MeHg and DOC over time for the entire area. The model formulation is known as a Model I formulation (Johnson and
Table 3 Scenario codes and assumptions. Scenarioa
Assumptions
Base
Concentrations of N, P, DOC and MeHg are allowed to increase ≤ 10%, respectively. Clear felling and fertilization are the only forestry operations assumed to affect the concentrations MeHg0.5 As scenario Base but the excess leakage of MeHg after harvest is reduced by 50% MeHgThinn As scenario Base – WsAll but the excess leakage of MeHg ha−1 is assumed to be the same after thinning as after final felling Unconstrained No constraints on concentration; otherwise as scenario Base a The scenario is applied either for each sub-catchment, denoted WsSep, or to the entire area, denoted WsAll.
Scheurman, 1977). In technical terms, a solution consists, for each stand i, i ∈ {1, …, I}, of the number of hectares that should be allocated to each treatment schedule j, j ∈ J(i), where J(i) is the set of treatment schedules pertaining to stand i. There are T periods and t ∈ {1, …, T}. Associated with each treatment schedule j for each stand i are the NPV ha−1, Rij, the harvesting volume ha−1 in period t, Hijt, the standing volume Vij ha−1 after harvest in the last period, an indicator Cijt set to 1 if the stand is clear felled in period t, otherwise 0, and an indicator Fijt set to 1 if the stand is fertilized in period t, otherwise 0. Each stand belongs to one of three sets, S(r), s ∈ {1, 2, 3} corresponding to the subcatchments. Given that the decision variable xij signifies the number of hectares of stand i managed according to treatment schedule j, the areas of final harvest and fertilization in period t in sub-catchment r (ctr and ftr, respectively) are given by ∑ ∑ Cijt ⋅xij = ctr
t = 1…; T; r = 1; …3
ð1Þ
∑ ∑ Fijt ⋅xij = ftr
t = 1…; T; r = 1; …3
ð2Þ
i∈Sðr Þ j∈J ðiÞ
i∈Sðr Þ j∈J ðiÞ
This means that the flux of water in period t from sub-catchment r, wtr, can be expressed as wtr = BWr + FW 0 −FW B ⋅ctr + FW 1 −FW B ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð3Þ
where BWr is the total flux of water from sub-catchment r under undisturbed conditions (base level), and FW0, FW1, and FWB are the flux of water ha−1 during the 5-year period when harvest is conducted, the flux one period after the harvest, and the base level flux, respectively (cf. Table 2). This means that the second term adds the additional flux above the base level following the harvest in period t and the third term gives the additional flux following the harvest in the previous period. Following the same logic, the flux of N in period t from sub-catchment r, ntr, can be given as: 0 B 1 B ntr = BNr + FN −FN ⋅ctr + FN −FN ⋅cðt−1Þr + FE ⋅ ftr t = 1…; T; r = 1; …3
ð4Þ
where BNr is the total flux of N from sub-catchment r under undisturbed conditions (base level), and FN0, FN1, and FNB are the flux of N ha−1 during the 5-year period when harvest is conducted, the flux one period after the harvest, and the base level flux, respectively. The fourth term in Eq. (4) is the additional flux stemming from the leakage of fertilizer, where FE is the flux of N from a fertilized ha. Subsequently, Eqs. (5)–(7) compute the total flux
288
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
in period t from sub-catchment r for P (ptr), MeHg (mtr), and DOC (dtr), respectively, through 0 B 1 B ptr = BPr + FP −FP ⋅ctr + FP −FP ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð5Þ
mtr = BMr + FM 0 −FM B ⋅ctr + FM1 −FMB ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð6Þ
dtr = BDr + FD0 −FDB ⋅ctr + FD1 −FDB ⋅cðt−1Þr t = 1…; T; r = 1; …3
ð7Þ
The constraint on concentration is then formulated such that the flux of the substance divided by the water flux, i.e. the concentration, in a particular sub-catchment and period, should be less than or equal to the base level concentration multiplied by the allowed relative increase in concentration, δ. For N, P, MeHg, and DOC, respectively, this is given as ntr = wtr ≤ ð1+δÞ⋅BNtr = BWtr
t = 1…; T; r = 1; …3
ð8Þ
ptr = wtr ≤ ð1+δÞ⋅BPtr = BWtr
t = 1…; T; r = 1; …3
ð9Þ
mtr = wtr ≤ ð1+δÞ⋅BMtr = BWtr
t = 1…; T; r = 1; …3
ð10Þ
dtr = wtr ≤ ð1+δÞ⋅BDtr = BWtr
t = 1…; T; r = 1; …3
ð11Þ
(A slight transformation of Eqs. (8)–(11) will yield a linear relationship.) The given formulation refers to the requirements for each sub-catchment. When the concentration levels relate to the whole catchment area, the fluxes of substances are simply summed over the sub-catchments. The final stock demand is expressed by requiring the standing volume to be at least 100 m3 ha−1, i.e., I
∑ ∑ Vij xij ≥100⋅A
ð12Þ
i = 1 j∈J ðiÞ
where A is the total forest area. The harvest volume is not allowed to decrease by more than 10% from one period to the next, i.e., I
I
∑ ∑ Hijt xij ≥ ∑ ∑ 0:9⋅Hijðt−1Þ xij
i = 1 j∈J ðiÞ
i = 1 j∈J ðiÞ
t = 2…; T
ð13Þ
Treatment schedules must be allocated to the whole area of each stand, i.e., ∑ xij = Ai
i = 1…; I
j∈J ðiÞ
ð14Þ
period is set to 10%. This threshold was based on the assumption that concentrations are currently close to the limits between Good and Moderate status, which is true for P at Balån (Löfgren et al., 2009a), thus a 10% increase would result in a status shift from Good to Moderate. The scenario was applied either for each sub-catchment, denoted as case WsSep, or to the entire area, denoted as case WsAll. A sensitivity analysis was also conducted, in which the excess N and P fluxes in the period of final felling were doubled during the first 5-year period and zeroed during the second period (cf. Table 2). However, the analysis gave the same result as the base case and is not reported here. To test the influence of concentration levels on NPV, the model was solved for a range of values of δ, in addition to the base case scenario. The effects of imposing no concentration constraints were also tested, corresponding to solving the problem with a large value for δ. The scenarios are defined in Table 3. 3. Results 3.1. Economic implications As shown in Fig. 2, economic losses increase rapidly as the strictness of the concentration limitations increases. With restriction levels of 6% and 5% for WsSep and WsAll, respectively, less than half the economic value of the forest remains. An increase of ca. 30% in concentrations must be allowed if maximum NPV is to be retained. The loss for case WsSep is greater than for case WsAll (illustrating the scale effect), in accordance with expectations since in the latter case a high concentration of a substance in one sub-catchment can be compensated by a low concentration in another sub-catchment. From Table 4 it is evident that the assumptions regarding MeHg have a very large impact on the economic result. If the flux of MeHg is reduced by 50% compared to the Base scenario, the NPV value increases by 21% and 41% for cases WsAll and WsSep, respectively. The assumption that thinning would have the same effect as final felling appears to be less important regarding MeHg. 3.2. Concentrations In the Base and MeHgThinn scenarios, MeHg is always the limiting substance; no other substance reaches the 10% excess level in these cases (Table 5). In the MeHg0.5 scenario, however, N becomes the limiting element. Both MeHg and N could reach values well above the 10% level in sub-catchments in case WsAll, since the constraint is valid only for the area as a whole. P is always at a low level, whereas in some periods DOC could reach or exceed the 10% level for some subcatchments in case WsAll. When MeHg is limiting, it is limiting for all sub-catchments and all 5-year periods, except the first two, in case WsSep. In case WsAll it is limiting for the whole area for most periods. WsAll
where Ai is the area of stand i and the x-variables must be nonnegative, i.e.,
WsSep
100% 80%
xij ≥ 0
i = 1…; I; j = 1…; J ðiÞ
ð16Þ 60%
Finally, the NPV is maximized, i.e. 40% I
Max Z1 = ∑ ∑ Rij xij i = 1 j∈J ðiÞ
ð15Þ
2.6. Scenarios
20% 0% 0% -20%
The base case scenario represents the NPV when the model is solved with a value for δ of 0.1, i.e. the maximum allowed increase in concentration for any substance above the reference level in any
10%
20%
30%
40%
50%
60%
δ
Fig. 2. The NPV relative to its maximum value for different maximum allowed relative concentration levels (δ) for cases WsAll and WsSep.
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
(a)
Table 4 The relative NPV of the different scenarios compared with scenario Base (= 100; g%).
289
I+II+III
I
II
III
0.120
WsAll MeHg0.5 MeHgThinn
121 88
0.115 0.110
WsSep MeHg0.5 MeHgThinn
0.105
141 90
0.100 0.095 0.090
The only period it is below 9% is the first period. When N is limiting (scenario MeHg0.5) the N level is at 10% in at least one sub-catchment for the 2nd to the 12th periods in case WsSep, and at the 10% level for the whole area in case WsAll for the 2nd to the 8th period (Fig. 3).
0.085 0.080 0.075 0.070
3.3. Forest management
(b)
P WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstrained MeHg WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstrained DOC WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstrained
I
II
7
9
11
13
15
17
19
I+II+III
I
II
III
0.51
Table 5 The highest concentration of a substance above the base level (%) found in any period measured in the sub-catchments separately (I, II and III) or in the area as a whole (I–III) for the different scenarios and the cases WsAll and WsSep. I–III
5
5-year period
The comparatively low site index of the area is reflected in the forestry activities of the solution to the Unconstrained scenario. Stands are finally felled after a rotation period of ca. 100 years and
N WsSep Base MeHg0.5 MeHgThinn WsAll Base MeHg0.5 MeHgThinn Unconstraineda
3
1
0.49 0.47
III
0.45 7.0 10.0 6.5
4.2 10.0 2.0
5.6 10.0 2.0
8.5 10.0 1.9
5.5 10.0 4.9 16.1
21.1 33.7 23.1 35.4
18.4 29.9 19.7 33.3
6.1 14.7 6.1 16.3
0.43 0.41 0.39 0.37 0.35 1
3
5
7
9
11
13
15
17
19
5-year period 0.3 0.7 0.3
0.8 1.9 2.0
1.0 1.5 2.0
1.0 1.5 1.9
0.7 1.1 0.6 1.1
3.2 5.5 3.6 5.5
2.8 5.0 2.4 4.5
0.8 1.2 0.6 1.5
10.0 2.7 10.0
10.0 3.3 10.0
10.0 2.8 10.0
10.0 2.6 10.0
10.0 2.7 10.0 25.4
52.2 8.7 57.2 64.4
45.9 8.2 48.8 68.2
15.1 3.1 15.1 27.6
Fig. 3. The concentrations for each sub-catchment I–III and the whole area with case WsAll: (a) MeHg (ng L−1) under scenario Base and (b) N (mg L−1) under scenario MeHg0.5.
they are not thinned more than once per tree generation (cf. 2.4 ha of final felling with 1.7 ha of thinning y−1; see Table 6). The final felling activity drops as a result of the 10% concentration constraints in the Base scenario; a change that is somewhat compensated for by increased thinning. Fertilization also decreases, but the operation is not limited by the N concentration threshold. Instead, it is coupled to the reduced harvest activity. In the MeHg0.5 scenario, final felling increases to the levels of the Unconstrained scenario. Fertilization is reduced (WsAll) or not affected (WsSep) compared with the Base scenarios, since N now becomes a limiting constraint. In the Table 6 The average ha−1 y−1 of different forestry operations over the 100 years planning horizon for the different scenarios and the cases WsSep and WsAll.
1.7 4.0 1.6
2.0 5.0 2.0
2.0 4.3 2.0
2.0 3.9 1.9
1.7 3.9 1.6 4.0
8.2 11.7 9.0 12.2
7.5 10.3 7.7 10.6
2.5 4.6 2.4 4.6
a Note that the solution to Unconstrained is equal for WsSep and WsAll and therefore only reported for one of the cases.
Case
WsSep
WsAll
Final Fertilization Thinning Final Fertilization Thinning felling felling BaseScenario MeHg0.5 MeHgThinn Unconstraineda a
1.5 2.4 1.5 2.4
0.8 0.8 0.9 1.3
1.9 1.8 0.0 1.7
1.4 2.2 1.4 2.4
Note that Unconstrained is equal for WsSep and WsAll.
0.5 0.8 0.3 1.3
1.8 1.4 0.0 1.7
290
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
Ha
(a) 10 9 8 7 6 5 4 3 2 1 0
I
II
III
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
5-year period
Ha
(b) 10 9 8 7 6 5 4 3 2 1 0
I
II
III
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
5-year period Fig. 4. The final felling area in the sub-catchment areas I, II and III area over time in scenario Base for (a) WsSep and (b) WsAll.
MeHgThinn scenario, thinning activities cease due to the elevated MeHg concentrations. The reduced area of final felling in the Base compared with the Unconstrained scenario roughly agrees with the drop in economic value (NPV) for case WsSep; there are ca. 65% declines in both variables. The drop in economic value is smaller in case WsAll, because the final felling can be allocated over time in a more profitable manner in WsAll than in WsSep, as illustrated by the markedly different patterns of harvest area allocation among sub-catchments in Fig. 4. The harvest patterns for WsSep and WsAll of the BaseScenario show that a greater volume can be extracted earlier in case WsAll than in WsSep, leading to a higher economic return (Fig. 5). 4. Discussion The results indicate that the requirements to maintain or improve water quality stipulated by the EU Water Framework Directive goals may influence forest management, both in how the forest is managed and the economic consequences. The results agree on a general level with Öhman et al. (2009), in that maintenance of water quality may
m3
WsSep
WsAll
5000 4500 4000 3500 3000 2500 2000 1500 1000 500 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
5-year period Fig. 5. The harvest volume over time for cases WsAll and WsSep of scenario Base.
have an important effect on the economic outcome of forestry if it is implemented as a constraining factor. According to the present study, the consequences for forest management would be changes to the timing and reductions in the extent of final felling. Additionally, N fertilization would be reduced, even though this activity is not constrained by the N concentration limitation but a consequence of the reduced area of final felling, since N fertilization is most profitable when it precedes harvest. The results also show that there is a spatial scale effect (see e.g. Fig. 3 and Table 5). This is to be expected for drainage basins like this, where the age structure of stands differs substantially amongst the sub-catchments and there is much to gain from allocating the forest operations differently over time. However, if criteria for water quality are implemented for small sub-catchments or even-aged large catchments, this will not be an uncommon situation. The results thus indicate that it is important to identify critical points in the stream system where the WFD concentration constraints may be exceeded and then formulate restrictions on concentrations for those target points, a procedure suggested by Öhman et al. (2009). The results are highly dependent on the assumptions regarding responses of MeHg fluxes to forest management, for two main reasons: clear-felling increases MeHg fluxes 3-fold, but N fluxes only 2-fold and the effect persists for two 5-year periods. This explains not only why MeHg is generally the limiting substance, but also its cyclical pattern with clear-felling. When MeHg flux is also assumed to be caused by thinning this treatment essentially disappears. It should be noted in this context that thinning is not assumed to increase water runoff, only the concentration and, hence, the MeHg flux, thus it contributes even more strongly than final felling to the concentration. The economic implications of discontinuing thinning are limited, because of the fairly low fertility of the Balån area. MeHg concentrations (range; 0.077–0.126 ng MeHg L−1 for all scenarios and sites) under the given assumptions are 2- to 6-fold lower than those measured in Balån (Sørensen et al., 2009). Since the constraints are formulated as relative concentrations, this discrepancy does not affect the general conclusions, but it could be important in practice if absolute concentrations are considered and threshold values implemented. The concentration levels are low compared with those found in 109 randomly selected headwater forest streams in the River Dalälven catchment in central Sweden by Löfgren (25- and 75percentiles; 0.16 and 0.49 ng L−1, respectively, unpublished data). For all those reasons sensitivity analyses testing the assumptions concerning MeHg would be valuable. Furthermore, some of the data that the assumptions regarding forestry effects on N and P are based upon were collected in the 1970s and 1980s (Löfgren et al., 2009b). Since then water quality has been identified as a vital issue for the forestry sector and recommendations to improve forestry operations in order to protect surface water quality have been issued. Thus, on average, the fluxes of these two substances may be overestimated. However, results of observations during the first two years after clear-felling with modern logging techniques at Balån (Löfgren et al., 2009b) indicate that the current model assumptions are relevant with regard to the short-term N and P losses for the first 5-year period. In the management model, it is implicitly assumed that all forest operations take place at the beginning of the 5-year period. If instead they were assumed to take place evenly over the entire period, the effects would be distributed over a slightly longer period, which in turn may make it slightly easier to avoid concentration peaks. Still, it is unlikely that this would have any great influence on the results. Furthermore, in practice forestry operations tend to concentrate harvest activities as this reduces costs of hauling machinery and road maintenance. In this study all considered substances have been assumed equally important and the same thresholds, in percentage increases have been set for all of them. An alternative approach would have been to
L.O. Eriksson et al. / Forest Policy and Economics 13 (2011) 284–291
judge the relative severity of excess concentrations of each substance, expressed as value functions, and weigh them accordingly. This would mean that the problem would be dealt with as a multiple criteria problem. This would be reasonable, as there are so many uncertainties and subjective judgments that have to be made in cases where the scientific data are insufficient. These uncertainties are reality for the water authorities, who have to take decisions on the ecological status in each water body, regardless of data quality. At present, the ecological status of 44% of the Swedish stream water bodies (n = 15563) is at risk of not achieving Good ecological status (Water Authorities, 2010). In many of those streams, forestry is the only anthropogenic impact. The selected concentration restrictions in this study are not connected to the Swedish surface water classification system (SEPA, 2007), which besides biological and hydromorphological parameters includes WFD constraints for only acidity (pH) and phosphorus. However, the classification system is being continuously developed, and there are demands from society to include more chemical parameters (such as N, DOC and MeHg) and to improve monitoring (Water Authorities, 2010). Hence, the selected concentration restrictions in this study are used as examples of the principle of using multiple indicator constraints as supplementary input variables for forest management. With the exception of DOC, the assumed concentration restriction of an additional 10% is low compared to the normal variation of these substances in pristine forests. Based on annual means of concentrations measured from 1996 to 2008 at three forested, non-managed integrated monitoring sites (Aneboda, Kindla and Gammtratten) the coefficient of variation (CV) varies between 16 and 35%, 31–38%, 9–12% and 17–37% for nitrogen, phosphorus, DOC and Hg (note not MeHg), respectively (Löfgren unpublished data). This large inter-annual variation is rarely taken into account by the water authorities due to lack of data, and the classifications are mostly based on a few single observations or short-term time series (Water Authorities, 2010). Hence, there are large uncertainties in the classifications, which do not take into account the temporal variations, and the risks of misclassifications are obvious, especially close to the limits between High and Good status or Good and Moderate status (Löfgren et al., 2009a). Exceeding any of these limits due to a 10% concentration increase could induce remedial actions from the water authorities. Hence, it seems unlikely that the water authorities would use the simulated concentration increases to impose restrictions on forest management in the studied sub-catchments unless the water quality was very close to the class limit between Good and Moderate status, which only applies to P (Löfgren et al., 2009a). The broad conclusion from this study is that the WFD may indeed affect forestry if the effects of forest operations on water quality are sufficiently strong. The more quantitative results must, however, be interpreted with the utmost caution. This is merely a first attempt at analyzing the diverse dimensions of water quality that are affected by forest management, and great uncertainties still remain regarding the cause and effect relationships. We also need to consider dimensions other than those addressed here, such as acidification, erosion of particles and biological effects. However, water quality protection will ultimately depend largely on how well the forest operations are planned and carried out in the field, and a model such as the one described here can help in the planning process. Acknowledgements We would like to thank Jakob Schelker for supplying data from the Balån catchment and Anu Hankala, both at SLU, for preparation of the maps. The research has been supported by the Kempe Foundations and the Swedish University of Agricultural sciences, SLU.
291
References Bishop, K., Allan, C., Bringmark, L., Garcia, E., Hellsten, S., Högbom, L., Johansson, K., Lomander, A., Meili, M., Munthe, J., Nilsson, M., Porvari, P., Skyllberg, U., Sørensen, R., Zetterberg, T., Åkerblom, S., 2009. The effects of forestry on Hg bioaccumulation in nemoral/boreal waters and recommendations for good silvicultural practice. Ambio 38 (7), 373–380. Ekö, P.M., 1985. A growth simulator for Swedish forests, based on data from the national forest survey. Report No. 16. Department of Silviculture, SLU. Umeå. (In Swedish). Eriksson, L.O., 1983. Timber class formation by cluster analysis. Swedish University of Agricultural Sciences, Department of Forest Technology. Report No. 157. Garpenberg Sweden. Hagner, O., 1990. Computer aided stand delineation and inventory based on satellite remote sensing. In: Sylvander, R., Poso, S. (Eds.), The Usability of Remote Sensing for Forest Inventory and Planning. : Report No. 4. Remote Sensing Laboratory, SLU. Umeå. Hoen, H.F., Eid, T., 1990. En modell for analyse av behandlingsstrategier for en skog ved bestandssimulering og lineær programmering (A model for analysis of treatment strategies for a forest by stand simulation and linear programming). Rapport/Norsk institutt for skogforskning – NISK, 90:9, Ås. . (In Norwegian). Johnson, K.N., Scheurman, H.L., 1977. Techniques for Prescribing Optimal Timber Harvest and Investment Under Different Objectives – Discussion and Synthesis: Forest Science, Monograph, 18 (Supplement to Number 1, 1 March 1977), pp. a0001-z0001(1). Kreutzweiser, D.P., Hazlett, P.W., Gunn, J.M., 2008. Logging impacts on the biogeochemistry of boreal forest soils and nutrient export to aquatic systems: a review. Environ. Rev. 16, 157–179. Laudon, H., Hedtjärn, J., Schelker, J., Bishop, K., Sørensen, R., Ågren, A., 2009. Response of dissolved organic carbon following forest harvesting in a boreal forest. Ambio 38 (7), 381–386. Löfgren, S., 2007. Conclusions from the workshop. In: Löfgren, S. (Ed.), How to estimate N and P losses from forestry in northern Sweden: KSLAT 146:24–30. ISSN 0023– 5350. Löfgren, S., Olsson, H., 1990. Input of nitrogen and phosphorus to running waters in the Swedish inland. Background document to Marine'90, Action plan against marine eutrophication. Swedish Environmental Protection Agency report 3692. 100 pp., (In Swedish). Löfgren, S., Kahlert, M., Johansson, M., Bergengren, J., 2009a. Classification of two Swedish forest streams in accordance with the EU Water Framework Directive. Ambio 38 (7), 394–400. Löfgren, S., Ring, E., von Brömssen, C., Sørensen, R., Högbom, L., 2009b. Short-term effects of clear-cutting on the water chemistry of two boreal streams in northern Sweden: a paired catchment study. Ambio 38 (7), 347–356. Nisbet, T., 2001. The role of forest management in controlling diffuse pollution in UK forestry. For. Ecol. Manage. 143, 215–226. Nurminen, T., Korpunen, H., Utrotningsitalo, J., 2006. Time consumption analysis of the mechanized cut-to-length harvesting system. Silva Fenn. 40 (2). Öhman, K., Seibert, J., Laudon, H., 2009. An approach for including consideration of stream water Dissolved Organic Carbon in long term forest planning. Ambio 38 (7), 387–394. Ollas, R., 1980. New yield functions for trees and stands. Forskningsstiftelsen Skogsarbeten, Ekonomi no 5. 4 pp. Stockholm. . (In Swedish). Piirainen, S., Finér, L., Mannerkoski, H., Starr, M., 2007. Carbon, nitrogen and phosphorus leaching after site preparation at a boreal forest clear-cut area. For. Ecol. Manage. 243, 10–18. Reese, H., Nilsson, M., et al., 2003. Countrywide estimates of forest variables using satellite data and field data from the National Forest Inventory. Ambio 32 (8), 542–548. Ring, E., Löfgren, S., Sandin, L., Högbom, L., Goedkoop, W., 2008. Forestry and water – a review. Skogforsk Redogörelse Nr 3 2008. . (in Swedish with English summary). SEPA, 2007. Status, potential and quality criteria for lakes, streams, coastal waters and estuaries – a handbook on how the quality criteria in water bodies can be determined and monitored. Swedish Environmental Protection Agency report 2007:4. 414 pp., (In Swedish). SEPA, 2008. Statute book of the Swedish Environmental Protection Agency NFS 2008:1 (regulations and recommendations on the classification of and environmental quality standards for surface waters). ISBN 1403–8234, 116 pp. (In Swedish). SNBF, 1994. Swedish forestry act. Swedish National Board of Forestry (SNBF), Jönköping, Sweden. Sørensen, R., Meili, M., Lambertsson, L., von Brömssen, C., Bishop, K., 2009. The effects of forest harvest operations on mercury and methylmercury in two boreal streams: relatively small changes in the first two years prior to site preparation. Ambio 38 (7), 364–372. SOU, 2007. Forest statistics – Annual report 2007. Swedish Forest Agency, Jönköping. SOU, 2009. Forest statistics – Annual report 2009. Swedish Forest Agency, Jönköping. http://www.svo.se/episerver4/default.aspx?id=37802. Swedish Forest Agency, 2000. Guidelines and foundations for the Forest Agency work on issues regarding forestry and water. Swedish Forest Agency Cirkulär 2000:5. 7 pp., (In Swedish). Swedish Forest Stewardship Council, 1998. http://www.fsc-sweden.org/Portals/0/Fsc-eng. pdf_1.pdf 1998. Water Authorities, 2010. In: Egerup, J., Eriksson, A., Göranson, M., Tornevall, H. (Eds.), Water – Our Common Responsibility: Vattenmyndigheterna, Tabergs tryckeri, March 2010, p. 30.