3.4 Activities in NZ - 6 Land reverting to natural vegetation

6.1 Specific activities considered, and effects not covered
6.2 Effects on non-CO2 greenhouse gases
6.3 Definitional issues
6.4 Socio-economic factors influencing the activities considered
6.5 Applicability of a land-based accounting system
6.6 What data, publications and reviews are available
6.7 Sampling techniques and issues
6.8 What process models are available to complement data
6.9 What regional and national scale inventories are available
6.10 What is known of 1990 carbon stocks and current trends
- Table 6.1. Estimated 1990 Carbon stocks and Recent Trends for Land Reverting to Natural Vegetation (Scrubland)
6.11 Projections for the first commitment period where feasible
6.12 Data and research needed to reduce uncertainties

6 Land reverting to natural vegetation

6.1 Specific activities considered, and effects not covered

We consider the natural reversion of pasture to unmanaged woody vegetation, commonly referred to as scrub. Reversion to other forms of natural vegetation are possible but are assumed to be negligible here because reversion to scrub is among the dominant features the New Zealand landscape.

Specific forms of scrub that may be considered as natural vegetation include unmanaged indigenous scrub, manuka/kanuka scrub or fern, subalpine scrub, gorse scrub, cassinia scrub, Dracophyllum scrub, matagouri, and sweet brier.

In assigning carbon stocks and fluxes to scrub vegetation and soils, we explicitly exclude the effects of herbivory and fire, while erosion is considered in section 7.4. Reversion of pasture to scrub may affect CH₄ uptake by soil. A summary of methane removal in soils is given in section 7.6.

6.2 Effects on non-CO2 greenhouse gases

Conversions of grazing land to scrub implies a reduction in animal (especially sheep) numbers that can be associated with reductions in livestock-related methane and nitrous oxide emissions. In addition, some evidence suggests that after 10-20 years, abandoned pastures occupied by woody vegetation may become methane sinks on a marginally significant scale in the Kyoto context (K.R Tate, unpublished data).

6.3 Definitional issues

At present, definition of scrub as a land-cover class and the areal extent of scrub vegetation types is somewhat ambiguous. The VCM mapping, during the period 1981 - 1987, defined scrub as woody vegetation not mapped as forest classes (Hall et al 1998; Tate et al 1997). Pure scrub (ca 1.1 Mha) and pasture-scrub associations (ca 5.1 Mha) were identified. An estimate of 3.7 Mha for 1990 scrub area (Tate et al, 2000b) has been based on the VCM mapping, but separating the pasture-scrub associations is subjective. The LCDB mapping, using 1996/98 satellite imagery, determined a scrub land-cover class, but the spatial correspondence between the two mappings is not as good for this land-cover class as for others. Scrub area determined in the LCDB is 2.67 Mha for 1996/7, however, this smaller area should not be taken as an indication that scrub area declined from the 1981-87 period - in fact the reverse is almost certainly true. Rather the difference indicates a need to improve methods used to define scrub areas and the potential exists to achieve this in the next generation of satellite based land-cover mappings.

Whether New Zealand's scrublands will fall within internationally accepted definitions of forests for the purposes of the Kyoto Protocol is a significant issue. This will depend on policy decisions which may consider appropriate levels of plant height, standing biomass density, biomass production and/or areal extent within a plot, to qualify land as a forest (SRLUCF 2000). In New Zealand's circumstances land reversion can often be considered as an early successional stage of reforestation, and in some cases is a deliberate action in this regard. If reversion to scrubland is treated as reforestation and meets the test of a "direct human-induced" activity then it can be covered under Article 3.3 of the Protocol. However, this will raise the issue of defining at what time abandoned pasture becomes scrub in relation to the "since 1990" condition in Article 3.3. Pre-1990 abandonment of land can still be considered under Article 3.4 in this case. Conversion to indigenous forest is not widespread at present but may become so - e.g. as a one-off sink establishment activity to avoid compliance issues surrounding harvested forest. The early stages of such a land-use change may be difficult to distinguish from reversion to scrub.

If reversion to scrubland is not treated as reforestation then the activity could still be treated under Article 3.4 and no "since 1990" condition applies. However, there is still a need to meet the requirement of being a "human-induced" activity.

Another definitional issues will be to determine the boundary between scrub (which we currently believe sequesters carbon from the atmosphere) and indigenous forests (which we assume to be neither losing nor gaining C). In practice, definitional boundaries would be problematic if they did not permit classification between New Zealand land-cover types based on satellite imagery.

The combination of the above-mentioned definitional issues could conceivably result in inclusion of scrub in all cases, some cases, or no cases.

6.4 Socio-economic factors influencing the activities considered

The abandonment of pastures and subsequent reversion to scrub has been driven primarily by the 1984 removal of fertiliser subsidies for agricultural land. Varying rates of land abandonment following the removal of fertiliser subsidies result largely from changes in commodity prices for sheep meat and coarse wool. However, reversion of unprofitable pastoral land to scrub is not universal: many abandoned agricultural lands have been converted to forestry if this land use is deemed profitable (see section 3). The possible introduction of carbon-emissions trading could affect future choices regarding abandonment or change in use of marginal pasture land. Use of scrub species as a biomass crop for energy production represents another possible factor in such changes, but that usage may be limited by requirements for access and environmental management during harvesting and by proximity to sites where the crop would be used.

6.5 Applicability of a land-based accounting system

No land-based system to capture land reversion exists currently, except for information contained in Tate et al (1997, 2000b), White et al (2000), VCM and LCDB. Carbon stock change estimates given here are based on scrub areas determined from VCM (ground-mapping). As noted in section 6.3, smaller areas are determined from satellite-based mapping (LCDB), and rates of change in scrub area have not been estimated directly. In future, development and application of satellite-based mapping technology could provide regular estimates of scrub area, particularly if classification schemes can be developed to separate reversion to scrub from other land cover changes. Land clearance is identified more easily than recruitment/ regeneration and on-going field measurements will be needed to calibrate remotely sensed estimates of biomass. (See also section 6.10.)

A land-based accounting system that does not cover all land may not fully account for changes in carbon stocks derived from erosion following land-use change. This occurs when the accounting system does not include the deposits of soil-derived sediments, which may be in river beds or coastal sediments. Erosion is considered further in section 7.4.

6.6 What data, publications and reviews are available

Several publications (Tate et al 1997, 2000b; White et al 2000; Scott et al 2000) provide initial estimates of rates of carbon accumulation in growing scrub. Additional work is currently underway to assign credibility to these estimates and to determine the magnitude and direction of soil carbon changes in scrublands.

6.7 Sampling techniques and issues

The primary method for estimating changes in stocks over time is the harvest of living biomass at intervals on a site or in a chronosequence of paired sites (e.g., Scott et al 2000). Similar methods can be employed for monitoring soil organic matter changes. These techniques generally require long time intervals (5 years or more) to accumulate accurately measurable changes between harvests. Remote sensing imagery presents a possibly improved method for large-scale analysis (e.g., White et al 2000), and specially the use of radar imagery.

There are significant problems in determining affected land areas in 1990 and subsequent changes, although there is a potential for using archived satellite imagery to resolve this. Problems are also expected in accounting for carbon in scrub with high levels of certainty, due to:

determination of carbon allocation to roots and root effects on soil C
effects of site nutrient status and different scrub types on rates of growth
spatial scaling of carbon stocks to the national scale for a 1990 baseline in a manner comparable to subsequent commitment periods.

6.8 What process models are available to complement data

The following models have been applied to New Zealand's scrubland ecosystems: 3-PG (White et al 2000), CASA (Tate et al 2000b), RHESSys (Band et al 1991). Efforts to couple a plant physiology model (3-PG; Landsberg and Waring 1997) to a soil biogeochemistry model are underway (Troy Baisden, pers. comm.)

6.9 What regional and national scale inventories are available

The Forest and Scrub CMS and Soil CMS provide a systematic basis for national scale inventories when used in conjunction with the VCM/LCDB classification of land use. Hall et al (1998) have estimated vegetation inventories for 1990. At present area estimates based on VCM and LCDB approaches differ, see Table 6.1, which may be due in part to some ambiguity in the definition of this land cover class. The LCDB approach has met targets for verification and is being refined further.

6.10 What is known of 1990 carbon stocks and current trends

Data in Tate et al (1997, 2000b) and Hall et al (1998) allow estimation of a national inventory, but not necessarily a 1990 baseline. Recently completed work from a MAF contract suggests that estimation of 1990 scrub and indigenous forest areas (and changes to 1996) from satellite imagery is feasible (Peter Stephens, Landcare Research, pers. comm.), but more work is still needed to obtain complete national coverage for 1990. White et al (2000) show that models can provide a scaling method for national-scale estimates of forest and scrub biomass, but this study did not elucidate methods for identifying biomass associated with land-use change.

Figure 6.1. Rates of carbon accumulation in one Manuka/Kanuka scrubland ecosystem, based on Scott et al (2000).

Figure 6.1. Rates of carbon accumulation in one Manuka/Kanuka scrubland ecosystem, based on Scott et al (2000).

Biomass carbon is assumed to be zero at time zero following fire, and soil carbon is assumed to have an average value. Soil carbon results for scrub growth may differ for scrub establishing on pasture rather than scrub regrowth after fire/clearing.

Rates of carbon accumulation in vegetation are calculated for specific sites in Scott et al (2000) and at a national scale in Tate et al (2000b). Figure 6.1 shows that rates of carbon accumulation in <35 year old manuka/kanuka scrub stands following clearance by fire are very rapid, but level off quickly with stand age above 35 years (Scott et al 2000). This may be a reasonable proxy for the rate of biomass accumulation in reverting pasture, although conditions, e.g. in relation to seed dispersal and species succession, may be different in the two situations.

Table 6.1. Estimated 1990 Carbon stocks and Recent Trends for Land Reverting to Natural Vegetation (Scrubland)

Estimated for 1990				Recent rates of change per year^a
Area (Mha)	Vegetative C stock (Tg C)	Soil C stock (Tg C)	Total C stock (Tg C)	Area (Kha)	Veg C Stock (Tg y^-1)	Soil C stock (Tg y^-1)	Total C stock (Tg y^-1)
2.6, 3.7¹	527²	393³	920	Uncertain⁴	0.6 to 3.4⁵	Uncertain	Uncertain

^aaverage change during 1990 and 1999.
¹ LCDB1 based figure (MAF 2001) is 2.67 Mha for the 1996/97 period, VCM based figures used in Tate et al (2000) were 3.7 Mha, see section 6.3
² Based on Hall et al (1998)
³ Davis et al (1999)
⁴ Uncertain due to changes in methods of accounting for each land system type.
⁵ from Tate et al (2000b)

Available estimates are summarised in Table 6.1. In general little is known about the national distribution of stand ages and rates of carbon accumulation in other scrub types. Based on assumptions for area affected and time to reach maximum biomass density, Tate et al (2000b) estimate the current rate of carbon accumulation in scrub vegetation to be 0.6-3.4 TgC y-1 by estimating the area which may be undergoing scrub growth on abandoned pasture at the rates found by Scott et al (2000). Rates of carbon accumulation or loss from soil are largely unknown, but based on 2 sites in Scott et al (2000), appear to be small compared to carbon accumulation in vegetation.

6.11 Projections for the first commitment period where feasible

Tate et al (2000b) estimate a likely range of ~0.6-3.4 TgC ^y-1 sequestration (based on Hall et al 1998); this estimate does not include any changes in soil carbon.

6.12 Data and research needed to reduce uncertainties

As discussed in section 6.3, a improving the definition of the scrub land-cover class and validating corresponding areas is necessary. This is prerequisite to determining the rates of change in scrub area and pre-existing land-cover, land-use, that would be required to estimate carbon stock changes.

For the purposes of understanding uncertainty, biomass carbon stocks and rates of carbon accumulation must be considered separately. Uncertainty in biomass carbon appears large based on direct measurements (Tate et al 2000b). However, models can provide estimates of scrub biomass which compare acceptably (R2 = 0.72 ) with measured sites at the national scale (White et al 2000). Therefore, White et al (2000) suggest that models can be used to derive acceptable estimates of scrub biomass at the national scale. Improved process-based models and characterisation of additional scrub sites will decrease levels of uncertainty.

In addition to studies following Scott et al's (2000) measurements of growth rates and below-ground allocation in regenerating scrub, several other critical sources of uncertainty must be addressed to permit scientifically credible assessment of scrub carbon sequestration. In addition to the issue of area uncertainty mentioned above, additional sources of uncertainty area associated with the age of scrub stands, their classification among scrub types, and associated changes in soil C. Additional uncertainty arises from inaccurate scrub boundaries resulting from either poor spatial resolution (VCM) or the need for better interpretation of satellite imagery (LCDB).

In general, the primary sources of uncertainty above relate to scaling methods. Of the potential terrestrial carbon sinks under consideration in this report (pasture, cropping, planted forest and reverting scrubland), scrub represents a large potential carbon sink that is most highly dependent on newly developed scaling methods to obtain credible proof of carbon accumulation. This is primarily because no agricultural surveys or industry statistics cover scrubland. Therefore, we expect estimation of carbon accumulation in pasture reverting to scrubland to be dependent on satellite-derived estimation of young scrubland. While the current LCDB has met targets for verification and established a benchmark, further work is required to achieve a better identification of scrubland - e.g. through use of topographic correction algorithms. Given the good coverage of ca 1990 satellite imagery (Peter Stephens, pers.comm), remote sensing appears capable of ascertaining any scrub carbon baseline information that may be required under Article 3.4.

Uncertainty in remotely sensed carbon accumulation estimates can be further reduced with improved models of scrub growth characteristics, and the filling of data gaps (especially soil carbon relating to soil type, land use and climate). Given that research is underway on scrub growth characteristics, and the filling of soil data gaps, scaling methods involving remote sensing are expected to remain the primary source of uncertainty in rates of scrub carbon accumulation.