3.3.1 Time horizon
We found that the discussion around time horizons (THs) can be classified into the following topics:
Literature discusses different types of THs, although it is often not clearly indicated out which type is meant. Saez de Bikuña et al. (
2018) differentiate three THs: the TH of the product or service (the length of the life cycle), the TH of the inventory modeling (which can be longer than the life cycle if emissions persist), and the TH of the impact modeling for characterization. According to Peters et al. (
2011), there can be separate THs for each impact category if they have different time scales. There is a question as to whether THs must be consistent. Yu et al. (
2018) differentiate between the TH of the assessment, defined in goal and scope, and the TH of the impact assessment. In their case study, they compare different THs for the assessment of road pavements and apply different THs for the time-adjusted Global Warming Potential (GWP, discussed in the characterization section). Almeida et al. (
2015) and Levasseur et al. (
2010) criticize inconsistencies if the stated TH of the assessment, defined in the goal and scope phase as temporal system barrier, does not fit to the TH of the characterization. For example, there would be inconsistency if the stated TH of an analysis is 100 years and the GWP100 is applied on a dynamic inventory. In a static LCA, all emissions would be summed up and handled as if they were emitted at the beginning of the lifetime considered. With a dynamic inventory, GWP applied to the emissions of year 99 would lead to an implicit TH of the assessment of 199 years. For Levasseur et al. (
2010), this is a reason to use dynamic characterization. Others do not regard different THs as inconsistent and accept them in LCA (Brandão et al.
2013; Brandão and Levasseur
2011; Menten et al.
2015). Beloin-Saint-Pierre et al. (
2016) apply the static GWP on a dynamic inventory in a way that emissions in the first year no longer count in the one hundred and first year, but the last emission is counted until the 199th year. Brandão and Levasseur (
2011) suggest fixed THs rather than fixed endpoints (certain years) unless the latter is attractive because policy often has fixed temporal goals. Assessments with fixed endpoints cannot be compared well with later LCAs on the same object. They would have a shorter timeframe, so the impact would be reduced or the previous LCAs would need to be recalculated with a new starting point. In most studies, different THs are not discussed and statements on THs concern the TH of the assessment.
2.
Importance of the TH for the LCA outcome
Case studies (Huijbregts et al.
2001) as well as methodological literature (Kendall
2012) show the high dependency of LCA results on the chosen TH. The dependency is stronger for short-time emissions, like those of methane in the GWP calculation (Levasseur et al.
2010) and land use change (Schwietzke et al.
2011), where initial emissions marginalize over time. For example, CH
4 emissions have a 72 times higher GWP than CO
2 in a 20-year TH, but only a 25 times higher GWP in a 100-year TH. Initial land use change in biofuel production raise the cumulative radiative forcing by 5% over a 50-year TH but only 2% over a 100-year TH. According to Schwietzke et al. (
2011), this may be decisive for corn ethanol, which is not much more beneficial than fossil fuel.
3.
Length of TH as a subjective decision
Generally, the length of THs is described as a subjective decision (Beloin-Saint-Pierre et al.
2016; Dyckhoff and Kasah
2014; Laratte et al.
2014) or even arbitrary (Brandão and Levasseur
2011). De Rosa et al. (
2017) state that a TH depends on the goal and scope of the study. According to Berntsen et al. (
2010), a TH should fit to political goals and Levasseur et al. (
2012) call it a political decision. Cherubini et al. (
2011) state that “tipping point issues and commitment periods and targets provide motivation for time-constrained assessment approaches”. Even if this is true, Guo and Murphy (
2012) propose showing results with other THs than those of political goals for more meaningful results. Several subjective reasons speak for short THs. Bakas et al. (
2015) state that long-term effects are often ignored because of complexity, diversity of approaches, and uncertainty together with short time preference of policymakers. According to Lebailly et al. (
2014), uncertainty about future mitigation technologies makes a short TH preferable. According to Fearnside (
2002), a long TH would marginalize short-term impacts, which would be counterintuitive. Brandão and Levasseur (
2011) say that the TH could be representative for the urgency of the environmental problem, with urgent problems having short THs. Brandão et al. (
2013) confirm this but admit that short THs would violate the principle of intergenerational equality. Herzog et al. (
2003) even say that what a society calls “permanent” is a political decision. Due to the subjectivity of THs, scenario analysis (Guo and Murphy
2012; Kendall and Price
2012; Udo de Haes et al.
1999) or sensitivity analysis (Beloin-Saint-Pierre et al.
2016) should be performed, so that the readers of an LCA study can decide on their own. For scenario analysis, Hu (
2018) suggests calculating the different THs for an assessment based on cultural theory. According to this theory, people’s attitudes to THs can be divided into four different categories. In cultural theory, summarized by Hofstetter et al. (
2000), societies can have the archetypes fatalist (with the shortest TH), individualist, hierarchist, and egalitarian (with the longest TH). Implications of cultural theory with suggestions for TH and discounting are discussed in Hauschild et al. (
2018). As an alternative to scenario analysis, Dyckhoff and Kasah (
2014) handle the TH as a dependent variable on a plot so that decision makers can decide and see intersections or break-even points at certain time points. Fearnside (
2001) states that the often applied TH of 100 years is adequate if one thinks of a 50-year-old decision maker who has a personal TH until the death of his grandchildren as the last family members he will personally know. Therefore, there is an opinion in literature, that TH is a policy decision and should be rather short so that it reflects political goals and that long THs could be abused as an excuse for not acting today. This leads to the question if THs should generally orientate on the current possibilities of acting against environmental impacts instead of measuring impact for current and future generations whereas we have no influence on the action of the latter. There may be a difference in thinking whether the LCA practitioner feels personally involved and wants to act in a foreseeable TH or whether he wants to measure impact with an intergenerational perspective as an outside observer.
4.
Derivation of THs by action orientation and possibilities for gaming
Although it is generally accepted that future generations should not be penalized, there are considerations that long THs marginalize effects of short-term actions, like carbon capture and storage (Almeida et al.
2015; Herzog et al.
2003). Especially an infinite TH would make every sequestration, if assessed with the GWP indicator, useless (Brandão et al.
2013), but would avoid problem shifting to the future (Lebailly et al.
2014). Berntsen et al. (
2010) say that if goals for impacts are set lower, the TH should become shorter because it is assumed that there shall be continuous improvements and so the target year for the reduced goals would come earlier. Nevertheless, action orientation in the TH identification can also lead to gaming incentives. A temporal cutoff encourages emission shifting to the end of the assessment period and temporal emission storage without consumption reduction (Brandão and Levasseur
2011). For example, a fixed TH leads to the incentive for carbon capture and storage to store the carbon until 1 year after the end of assessment (Brandão et al.
2013). According to Fearnside (
2001), because of the risk of gaming, an assessment should consider actions of actors and possible innovations in this field after the TH. Action orientation can also lead to counterintuitive measures. If new artificial systems (e.g., fast growing biomass) proved to be more efficient in binding CO
2 than natural systems, long THs would lead to the conclusion that nature (forests) should be removed. This would also depend on the inclusion and assessment of other impact categories (Soimakallio et al.
2015).
5.
Derivation of TH by measurement orientation
Instead of thinking of incentives for actions, THs can also be derived from requirements for exact measurements. Measuring dynamics in the inventory would be useless if the TH is infinite (Dyckhoff and Kasah
2014). Udo de Haes et al. (
1999) stated that it is sufficient if a TH includes most of the impact, so that in some cases 500 years or 10,000 years would be enough for radioactivity concerns. According to Brandão et al. (
2013), the TH should be long enough so that the assessed impact is no longer relevant. The TH could be calculated based on reaching a steady state in fate models for emissions, as in the case of leakage from landfills (Finnveden
1999). Kirkinen et al. (
2008) say that THs should be oriented on assumed turning points in nature, e.g., for global warming, and therefore be rather short. De Rosa et al. (
2017) say that the TH must fit the used indicators. For example, if sea level rise was applied as an indicator for global warming, a short TH would be inappropriate. Almeida et al. (
2015) chose certain THs because they are necessary for the GWP metric. However, Berntsen et al. (
2010) state that “the target determines the metric and not the other way around; one should first define the target, and then choose an appropriate metric and a consistent time horizon.” Nevertheless, the metric plays a role only if static LCA is applied. Then short THs could underestimate impacts and long THs could overestimate impacts because long-term emissions are summed at the beginning. Small amounts of emissions distributed over a long time can have a much lower impact than the same amount over a short time (Kendall
2012). Therefore, if long THs are set, a dynamic LCA should be applied. This is also emphasized by Bakas et al. (
2015) who show on the example of long-term heavy metal leaches from landfills that a static LCA significantly overestimates the actual toxicity of those emissions. Mallapragada and Mignone (
2017) consider it useful to apply fixed THs for midpoint indicators and fixed endpoints (in time) for endpoint indicators. A certain TH is required if long- and short-term emissions are assessed together; otherwise, the long-term emissions would marginalize those which are short-term in infinite or long THs.
An interesting approach for LCA is the “time-dominance principle” (Dyckhoff and Kasah
2014). If there is a decision to be made between some alternatives and some of them are less favorable within any TH than others, these alternatives are called “dominated.” Therefore, TH considerations must not be implemented for the dominated ones. The decision problem is reduced to the remaining alternatives. Hauschild et al. (
2008) apply two THs: 100 years and after 100 years. They assess emissions of the first 100 years, as is commonly done, and introduce a new indicator “stored toxicity” which can also be applied to other impact categories.
Reasons for short and long THs are summarized in Table
1.
6.
Equivalence of TH and discounting
Table 1
Reasons for short and long TH of the assessment
• Long THs can overestimate impacts | • Short THs can underestimate impacts |
• Long THs make short term actions less favorable | • Short THs place weighting on short-term impact categories |
• Long THs often ignored because of higher complexity and uncertainty | • Appreciate intergenerational equality |
• General short time preference of people | • Short THs lead to wrong incentives in year X + 1 ➔ gaming |
• Long THs ignore urgency of problems, excuse non-action | • Avoids problem shifting to the future |
• Possibility of future mitigation technology | |
• THs should fit political goals which are usually short-term | |
For many authors, the discussions around adequate THs for assessments and adequate discount rates are equivalent, e.g., see Boucher (
2012) or Almeida et al. (
2015). For Fearnside et al. (
2000) and Hu (
2018), THs and discounting are both time preference schemes. Mallapragada and Mignone (
2017) state that a TH is only a substitute for explicit discounting. For Fearnside et al. (
2000), they are equivalent, but the TH is easier to explain to the public.
3.3.2 Temporal weighting/discounting
“Discounting is the mechanism by which a value for time is normally translated into economic decision-making” (Fearnside et al.
2000). For different reasons, discounting is also proposed in LCA. Fearnside (
2002) explains that discussions about applying discounting for the global warming assessment began already in the early 1990s but the IPCC instead adopted the procedure of time horizons. Fearnside (
2002) criticizes that the THs suggested by the IPCC, 25, 100, and 500 years, are unrealistic options because 25 years would obviously be too short and 500 years too long, meaning most people chose the 100-year TH. We classified findings on discounting as follows:
1.
Subjectivity of discounting
Similar to our findings on THs, discount functions and rates are regarded as a non-consensual (Almeida et al.
2015), value-laden (Brandão and Levasseur
2011), and ethical (Levasseur et al.
2012) choice. As LCA is a value-based decision instrument, it should apply discounting (Hellweg et al.
2003), which would make it a more “business-like” decision instrument (Yuan et al.
2015). Another term for expressing subjectivity is “time preference.” Many authors claim that time preference is the main reason for discounting (Levine et al.
2007; O’Hare et al.
2009; Fearnside et al.
2000). Fearnside (
2002) explains time preference with the mortality of people. According to Hellweg et al. (
2003), there is ethical consensus that discounting based purely on a time preference is immoral and should not be applied in LCA, but can be regarded by decision makers. Fearnside et al. (
2000) suggest the term “immediate emission equivalent” instead of discounted emission because the term “discounting” would describe the calculation of the net present value of money. They further describe “distortions” in the assessment if discounting is disregarded over THs longer than 100 years. Generally, a decision about discounting in LCA cannot be avoided, as a decision for a discount rate of zero would also be a decision that one would have to make, which should be justified in the same way a decision for any other rate would be (Fearnside et al.
2000). For the calculation of discount rates, Bakas et al. (
2015) proposed applying different scenarios based on different sociocultural types of people. In the cited cultural theory (Hofstetter et al.
2000), there are the already discussed four different archetypes of people with different perceptions of time or risk, which can result in different discount rates for them. Cultural theory is already applied by impact assessment tools like Recipe and Eco-Indicator99 for setting different THs and weighting schemes according to the different cultural perspectives but they are not applied to discounting there (Hauschild et al.
2018). The reliance on different perceptions is also assumed by Field et al. (
2001), who state that decisions on discount rates should be made at the end by the decision maker and not already by the LCA maker. As explained in the TH section, it is also valid for discounting that delaying emissions only makes sense if there is a value for time (Brandão and Levasseur
2011). Due to the inherent subjectivity, scenario or sensitivity analysis is suggested by many authors (Herzog et al.
2003; Weitzman
1998; Yuan et al.
2015).
2.
Importance of an accurate discount rate
Yuan et al. (
2015) describe a high influence of the discount rate on the outcome of an LCA. Even small rates around 1% marginalize impacts over just a few decades (Fearnside
2002). Bakas et al. (
2015) propose a small rate near 0% as the rate is very decisive. According to Yuan et al. (
2009), discounting should be handled very conservatively because an underestimation of impacts would be more critical than an overestimation.
3.
Discounting for monetary reasons
One obvious reason for discounting is that environmental damages or their prevention cause costs, and costs can be discounted. Even if there is currently no reliable monetization mechanism available for certain emissions, assuming that impact and monetary value are congruent, impact can be discounted in the same way as the monetary value (Kendall et al.
2009; Wang et al.
2018). The discount rate depends on the overall change in wealth and the change of the marginal utility of that change (Fearnside et al.
2000) and inflation (O’Hare et al.
2009; Levine et al.
2007). The latter authors assume that monetized damage will be reduced in the future as a result of new technologies to cope with it. The social discount rate, a measure used in cost-benefit analysis that also assumes economic growth, can also be applied in LCA (Richards
1997; Wang et al.
2018). This means that discounting is not only a temporal but also a spatial problem, as economic growth is not evenly distributed over the world. Fearnside (
2002) criticizes this by saying that discounting should not be applied to permanent losses, like losses in biodiversity. According to Hellweg et al. (
2003), discounting with monetized damages assumes that future generations would be satisfied with monetary compensation, which is not necessarily true. They state that the discount rate depends on opportunity costs that originate from alternatives and economic growth and that it can become negative in scenarios with economic decline.
4.
Time-dependent discount rates
For several reasons, a declining discount rate over time can be assumed. Boucher (
2012) discusses a discount rate declining from 3.5 to 1% after 30 years because of the change from individual to intergenerational discounting after that time and due to growing uncertainty. Weitzman (
1998) states that exponential discounting would not reflect the real opinion of people regarding the weight of future emissions from a certain point of time on. For example, an event in 300 years would not be less important than an event in 400 years. He also proposes a declining discount rate. For long-term impacts, different discount rates should be calculated, with the lowest discount rate being applied as the final discount rate. Fearnside (
2002) introduces a “generation weighted index” where the discount rate declines after every generation for four generations. According to Richards (
1997), the change of marginal damage by the emissions must be taken into consideration. Rising marginal damage should lead to a declining discount rate over time. The mixing of damage functions and discounting, i.e., physical discounting, is rejected by others.
Hellweg et al. (
2005) also apply discounting as a measure of dynamic characterization and normalization. The discount rate is derived from the assumed changing background concentration. In contrast, for O’Hare et al. (
2009), “the discounting model applies to costs and benefits, not to physical phenomena that generate them, unless their economic value is otherwise stable over time.” They recommend not discounting physical quantities. Brandão and Levasseur (
2011) differentiate between discounting for economic and social reasons and physical discounting on which the TH is based on.
6.
Chances, risks and uncertainty
For Udo de Haes et al. (
1999) and Yuan et al. (
2009), the uncertainty whether future impacts will happen is a factor for discounting. This can be influenced by technological advances or simply by a reduced life cycle, e.g., due to accidents (Yuan et al.
2009). Contrary to that, according to Hellweg et al. (
2003), discounting because of uncertainty should be avoided and uncertainty should be considered in the “damage prediction” (characterization). Uncertainty can lead to higher or lower discount rates, depending on uncertain positive or negative effects.
3.3.3 Temporal resolution of the inventory
A lot of information on the modeling of a dynamic inventory (DI) based on natural scientific phenomena, like growth of biomass, spatial distribution of emissions over time, or fate mechanisms of chemicals in nature, can be found in literature. Most of these are excluded for this review because we are focused only on basic principles.
The basic principle, the “dynamic” or “time-dependent” LCI, is defined by Tiruta-Barna et al. (
2016) as applying a sort of future prediction or higher temporal resolution. A dynamic LCI is a premise for exact discounting (Yuan et al.
2015) and advanced indicators for dynamic characterization. For example, daytime is relevant for the assessment of albedo effects in global warming (Almeida et al.
2015). LCAs derived from static models have a temporal bias (Guo and Murphy
2012), e.g., because simple accumulation of emissions data from small emissions over long THs lead to an impact overestimation (Bakas et al.
2015). The temporal resolution can have a decisive influence on the outcome of an LCA (Beloin-Saint-Pierre et al.
2016). Shimako et al. (
2018) show that the assessment of toxicity is highly influenced by the temporal resolution while it does not play a role in the assessment of global warming. A dynamic LCI can include temporal fate mechanisms of emissions (Herrchen
1998; Yuan et al.
2015; Lebailly et al.
2014; Shimako et al.
2017). Furthermore, it is needed if, rather than a single product, a fleet of products is assessed because that fleet rises after market entrance and will eventually decline later (Levine et al.
2007). As it can become very time consuming to build a dynamic LCI, literature proposes some simplifications. According to Beloin-Saint-Pierre et al. (
2016), it is not necessary that the whole inventory be dynamic, especially if there are many small emissions. In their article, they modeled 85% of the inventory in a dynamic way and positioned the remaining emissions at the starting point so that their effects are assessed over the whole life cycle, which can be considered a conservative approach. Hu (
2018) did not calculate a completely steady temporal LCI but divided the life cycle into different sequences like construction, operation, renovation, and second operation, so that the temporal resolution is low and given by the sequences, but the resulting LCA is still more accurate than a static LCA. Temporally differentiated data often cannot be predicted exactly, but can be modeled by mathematical functions with time dependencies. This may be recommended if there is a direct physical correlation between the functional unit and the inventory (Moura Costa and Wilson
2000; Huijbregts et al.
2001). Statistical functions can be applied to assess uncertainty, such as product failures or accidents (Field et al.
2001; Wang et al.
2018). Another proposed simplification is the “time resolved LCA” where the LCI is not measured but simulated based on historical data or simple simulation. Temporal effects can be regarded without the need for complex measurements (Zimmermann et al.
2014). In more complex cases, a scenario analysis of the LCI can be carried out (Zimmermann et al.
2015). More complex prediction methods are suggested by Su et al. (
2017): demand-supply models, complex adaptive systems, and Markov chains for short-term issues, and scenario analysis, query, investigation, and estimation for long-term issues, literature review, and simulation of user behavior. A modeling approach based on possible paths a product can take during its life cycle is introduced by Yuan et al. (
2009). During production, a product can take fast paths under optimal circumstances or slow paths when there are errors or loops due to inferior quality. The path approach is also chosen by Beloin-Saint-Pierre et al. (
2014) in their “enhanced structural path analysis.” It is used to build a temporal differentiated LCI without creating much data because the same processes are reused multiple times in different process steps or products. Such an LCI would be based on processes and not only on (intermediate) product data, which would reduce the overall data requirement. Nevertheless, reducing the complexity of a dynamic LCI remains difficult. There are interactions in process and supply chain dynamics as well as loop paths. Tiruta-Barna et al. (
2016) define two main challenges. First, foreground processes can be modeled or measured precisely, but background processes in the supply chain lack certain data and adequate temporal assessment. There must be a supply model between foreground and background processes that models the supply schedule and existing delays between the processes. Second, finding the best trade-off between accuracy and feasibility can be challenging. They propose an algorithm for calculating dynamic LCI based on a process and supply network. In the face of all these complexity, Almeida et al. (
2015) question the usefulness of a dynamic LCI because it is complex, not fully available in LCA software, and harder to communicate than a static approach. Furthermore, it does not solve temporal issues like the TH/cut-off issue. This criticism shows that temporal issues must be addressed in an integrated way. For example, Tiruta-Barna et al. (
2016) criticize that the methodological literature generally addresses either the dynamic inventory or the dynamic characterization issue but rarely both, suggesting that links remain missing.
3.3.4 Time-dependent characterization
For Collinge et al. (
2013), the inventory is a model of the “technosphere” while the characterization models the “biosphere” affected by it. Laratte et al. (
2014) call an LCA with a dynamic LCI but static characterization factors (CF) a “partly dynamic LCA.” The use of dynamic CFs make it a “fully dynamic LCA,” which is preferred because of its higher accuracy. This is confirmed by Menten et al. (
2015) in a comparative analysis. According to Levasseur et al. (
2013), it would be inconsistent not to apply dynamic CF on a dynamic LCI because emissions near the end of the TH would have a lower weighting than earlier emissions, but the inconsistency issue is very dependent on the interpretation of what exactly is meant by the TH, as explained in the TH section above. For example, if the TH of the whole assessments ends 100 years after the first emission, then only this first emission is fully weighted by a GWP100 and later emissions would have to be cut. If the TH ends 100 years after the last emission, then no emission would have to be cut when this static indicator is applied. Either way, dynamic CFs are dependent on the TH of the assessment because in a short TH less impact can happen, depending on the impact category (Guo and Murphy
2012; Lebailly et al.
2014; Collinge et al.
2013). According to Bakas et al. (
2015), such time-dependent CFs do not challenge the principle of intergenerational equality, only the setting of TH and discounting do. Conversely, dynamic characterization can be used to set the TH after which only little impact remains (Levasseur et al.
2013). We identified two main reasons for dynamic characterization. First, as most impacts depend on a certain concentration, which is not only defined by the assessed emissions but also by the background concentration, a changing background concentration over time must be regarded (Bakas et al.
2015; Collinge et al.
2013). The dependency on background concentration can be non-linear, e.g., because of certain thresholds (Hellweg et al.
2005), but there can also be an impact limit through saturation (Herzog et al.
2003). This leads to the second reason for dynamic CFs, the changing sensitivity of the ecosystem over time (Almeida et al.
2015; Bakas et al.
2015; Collinge et al.
2013). Hellweg et al. (
2005) mention a threshold-weighting factor depending on the “magnitude of exceedance of no-effect levels and, in case of dynamic modeling, on the time period of exceedance”. Bakas et al. (
2015) state that because of the changing ecosystem, not only the inventory but also the impacted system should be modeled dynamically. This ecosystem includes people. The changing distribution of exposed populations, such as the age distribution, should be included in a dynamic model (Collinge et al.
2013).
Regardless, dynamic CFs are currently not available for most impact categories. According to Collinge et al. (
2013), there are dynamic CFs for global warming and photochemical ozone and, under local premises, also for eutrophication, acidification, and ozone depletion. Van Zelm et al. (
2007) calculate a dynamic CF for acidification and Schwietzke et al. (
2011) discuss the cumulative radiative forcing as a dynamic CF instead of GWP for global warming. As in the other temporal issues, there are strategies for simplification of dynamic characterization. For delaying global warming, there is the ton-year approach that declares how many tons of CO
2 are avoided over a certain number of years compared with an alternative (Fearnside et al.
2000), but this cannot really be considered a dynamic impact measurement (Moura Costa and Wilson
2000). Bakas et al. (
2015) and Hauschild et al. (
2008) propose “stored toxicity,” or generalized stored impact, for emissions exceeding 100 years or a certain TH and assign them an indirect weighting in relation to the value of the corresponding short-term indicator. Kendall et al. (
2009) introduce a time correction factor that has to be multiplied with GWP and gives early emissions a higher weight than emissions near the TH of assessment. The kinetics of the emissions must be considered as well, particularly when materials change due to external effects, e.g., the generation of metabolites with an increased or reduced toxic effect. This would favor carbon capture and storage measures. Kendall (
2012) calls the same idea “time adjusted global warming potential”. Nevertheless, Kendall and Price (
2012) suggest a fully dynamic CF like cumulative global warming instead of a time adjusted GWP. The difference between a CF that is only dependent on the TH and a fully dynamic CF as a function of the inventory, impact mechanism, and the time axis is also explained by Shimako et al. (
2017), who prefer the latter due to its higher accuracy. Another index that must be multiplied with the GWP is the GWPbio (Cherubini et al.
2011). It is introduced as a CF for assessing biogenic CO
2 emissions, which contribute to the global warming because of atmospheric decay, depending on the rotation period of the biomass species.