Birgit Müller, Anja Linstädter, Karin Frank, Michael Bollig, and Christian Wissel. 2007. Learning from local knowledge: modeling the pastoral-nomadic range management of the Himba, Namibia. Ecological Applications 17:1857–1875.

Recurrence of Table 7 in the manuscript: Adjustment of perennial ground cover c(t) for each cell at time t dependent upon previous ground cover states (0,1,2,3) c(t-1), current precipitation, grazing pressure (including time of grazing).

Ground cover state 0 = "none", i.e., the perennial ground cover of the Stipagrostis uniplumis individuals in the grid cells is 0%. The perennial ground cover relates to the biomass cover of living parts of the tufts. It is therefore a measure for the reserve biomass of the grid cell’s grass population. If the ground cover state is zero, the grass population is extinct in the particular grid cell. This may be the case (i) because the cover state has decreased by one or more digits due to recent damage (disturbance) of the grass layer caused by grazing and/or rainfall below average, or (ii) because the cover state has remained unchanged due to conditions unfavorable enough for a significant change in ground cover state.

Ground cover state 1 = "little", i.e., the perennial ground cover of the Stipagrostis uniplumis individuals in the grid cells is 1–30% on deep soil or 1–15% on shallow soil. The ground cover may be "little" (i) because the cover state has gone down by one or more digits due to a recent damage (disturbance) of the grass layer caused by grazing and/or rainfall below average, or (ii) because the cover state has remained unchanged due to conditions not favourable or unfavorable enough for a significant change in ground cover state, or (iii) because the ground cover state has improved by one digit under favorable conditions due to a recolonization from the seed bank.  

Ground cover state 2 = "middle", i.e., the perennial ground cover of the Stipagrostis uniplumis individuals in the grid cells is 31–60% on deep soil or 16–30% on shallow soil. The ground cover may be "middle" (i) because the cover state has gone down by one or more digits due to recent damage (disturbance) of the grass layer caused by grazing and/or rainfall below average, or (ii) because the cover state has remained unchanged due to conditions not favorable or unfavorable enough for a significant change in ground cover state, or (iii) because the ground cover state has improved by one digit under favorable conditions due to a regeneration of the grass biomass (and facultatively due to a recruitment event).  

Ground cover state 3 = "high", i.e., the perennial ground cover of the Stipagrostis uniplumis individuals in the grid cells is 61–90% on deep soil or 31–45% on shallow soil. The ground cover may be "high" (i) because the cover state has remained unchanged due to conditions not unfavorable enough for a significant change in ground cover state, or (ii) because the ground cover state has improved by one digit under favorable conditions due to a regeneration of the grass biomass (and facultatively due to a recruitment event).

The transition rules indicating changes in ground cover state are backed up by monitoring data (Schulte 2002a, Schulte 2002b and  Linstädter née Schulte, unpublished data) and by expert knowledge based on ten years of continuous research on grass layer dynamics under different rainfall and land use conditions in Kaokoland. The sources for the particular information are indicated below, and an ecological explanation of the values is given.

Rules for "no" grazing

The transition rules where "no" grazing pressure is present can be substantiated by information from grazing exclosures established in 1995, 1996, or 2004 in northern Kaokoland and monitored till 2005. The year 1995 had above average rainfall, 1996 had average rainfall, and 2004 had below-average rainfall. Further information comes from monitoring sites established in 2003 and 2004 in protected areas of Namibia ’s Kunene Region, notably from a Mopane savanna in Etosha National Park (Jula Zimmermann, personal communication).

1. Regeneration: In the case of good (above average) rainfall, the grass layer may regenerate. In our model, the ground cover state can only improve by one digit per year. A faster inter-annual regeneration is not corroborated by our data (see Schulte 2002b), possibly due to biological restrictions of maximum productivity. The transition rules suggest that, under optimum conditions, the process of full regeneration takes four years. This seems realistic if recovery speed on grazing exclosures is extrapolated from good rain years.

2. No change: In most years with average rainfall (not more than 25% less than the long-term mean) or rainfall 25–50 % below the long-term mean, the ground cover state remains unchanged. The only exception is the regeneration of a locally extinct population even under average rainfall conditions due to a recolonization from the soil seed bank. This is backed by data from grazing exclosures Established in 1996. In the case of a "drought" year (defined here as rainfall more than 50% below the long-term mean), the ground cover state is reduced by one digit, if a reduction is still possible. 

Rules for "moderate" and "heavy" grazing

The transition rules for "moderate" and "heavy" grazing pressure are based on continuous monitoring data from 1995–2005 of regular rainy season pastures (RSP) and dry season pastures (DSP) in Kaokoland both with deep and shallow soil. Furthermore, data from grazing exclosures opened after four and eight years were used.

1. Regeneration: On DSP, a recolonization of pasture without living Stipagrostis uniplumis individuals is possible in years with good rain, both under moderate and heavy grazing pressure, because cattle can rely on the high amount of annual grass biomass. They are often moved to the next DSP before they have grazed all standing biomass. In good years, a further regeneration of a severely damaged or young grass population (ground cover state 1) can take place on these pastures under moderate grazing conditions. If the grass population of a rainy season pasture (RSP) grid cell has become extinct, it can only regenerate in years with good rain and moderate grazing pressure. Heavy grazing would destroy the freshly established perennial grass individuals.

2. No change: On DSP and RSP, a broad range of condition exists both under moderate and heavy grazing pressure where the local population is kept in its previous condition. For locally extinct populations, no regeneration can be observed under average and below-average rainfall, because the primary productivity of annual grasses and forbs is too low to hide Stipagrostis uniplumis seedlings. On DSP, in years with good and average rain and moderate grazing (for heavy grazing, only in good rain years) there is enough annual and perennial biomass present to prevent the Stipagrostis u. individuals from pronounced damage.

3. Pronounced damage: Pronounced damage can be related to an unfavorable combination of rainfall and grazing pressure. Because DSPs are used outside of the vegetation period, pronounced damage is only observed in years with rains below average (including drought years). Additionally, under average rainfall damage can occur if the biomass proportion of the perennial grasses is high (cover states 2 and 3) and grazing pressure is heavy. Here the readily available parts of the perennial tufts do not provide enough fodder for the whole year, and the cattle use lower, hard, and rather unpalatable parts of the tufts, reducing the reserve biomass. On RSP, the grazing pressure all year through prevents the perennial grasses from a successful compensatory regrowth during the vegetation period. Here grazing affects the ground cover and vitality of local population in all years except those with above-average rainfall. The only difference between moderate and heavy grazing is how severe the damage to the grass tufts is. Under heavy grazing, the local populations become extinct within only one year if rain is below average. Young or severely damaged populations are also destroyed in years with average rain. 

LITERATURE CITED

Schulte, A. 2002a. Stabilität oder Zerstörung? Veränderungen der Vegetation des Kaokolandes unter pastoralnomadischer Nutzung. Kölner Geographische Arbeiten 77:101–118.

Schulte, A. 2002b. Weideökologie des Kaokolandes. Struktur und Dynamik einer Mopane-Savanne unter pastoralnomadischer Nutzung. PhD thesis, University Cologne, Germany.