Soil Loss, Vegetation Recovery, and Sediment Yield Following Plantation Harvesting, Coromandel

Soil loss, vegetation recovery, and sediment delivery to streams following plantation harvesting, Coromandel

Chris Phillips1; Michael Marden2; Donna Rowan2 Loretta Garrett3 1 Landcare Research, Box 69, Lincoln, New Zealand, ph: 03 325 6700; fax: 03 325 2418,; e-mail: [email protected] 2 Landcare Research, Box 445, Gisborne, New Zealand, ph: 06 863 1345; fax: 06 863 1346; e-mail: [email protected] 3 Forest Research

Abstract On-site forest-related activities, such as roading and harvesting, together with off-site impacts of localised flooding, and sediment input into estuaries downstream have been issues of concern to local residents, regional authorities, recreationists and, the forest owners since before harvesting began of forests on the Coromandel in the late 1980s and early 1990s. A unique combination of weathered volcanic soils, steep slopes, together with a climate characterised by a history of frequent and, damaging high intensity rainfall events pre-disposes slopes erosion. Streams have a high incidence of flood flows and the potential to deliver significant volumes of sediment to coastal estuaries. The return period for a rainfall event of 133 mm in 24 hrs is estimated to be only 2 years (New Zealand Meteorological Service 1980).

On a harvested setting, sediment is generated both as a consequence of harvesting practice (e.g., scalping during hauler-logging) and, by post-harvest erosion processes including rain-drop impact, sheetwash erosion, rilling and by storm-initiated landslides. Process-based sediment generation rates and total sediment volumes have not previously been quantified for exotic forest cutover in a weathered volcanic terrain in New Zealand. In addition, few studies in New Zealand have attempted to quantify the relative contribution of sediment generated by these different processes to stream channels and, hence to stream sediment yield.

The objectives of this research were to assess vegetation recovery and quantify sediment generation rates from hauler logging at Whangapoua Forest, and to assess the relative significance of sediment delivery from the various sources to the stream channels. Finally, the sediment delivery ratio, i.e. the ratio between what is generated and what is exported as sediment yield, will be calculated. This paper reports results on sediment generation and vegetation recovery and catchment sediment yield.

A ground-based, site-disturbance survey of one compartment was used to identify the proportion of the logged setting occupied by different site disturbance classes. Field-based bounded plots (1m2 (n= 22) and 9m2 (n= 9)) stratified into these disturbance classes were used to measure rates of sediment generation and post-harvest vegetation recovery. Plot-based rates of sediment generation were applied to the cumulative area occupied by shallow-and deep-disturbance sites across the harvested area to quantify the sediment volume generated.

Analysis of post-harvest vegetation recovery and corresponding sediment generation during immediate post-harvest periods throughout Whangapoua Forest helped us to understand the relative significance and persistence of disturbed sites as a source of sediment and, in particular, enabled us to assess the significance of sheetwash erosion as a sediment-generating process. Post-harvest vegetation recovery peaked 9-months after the completion of harvesting. It was quickest on sites of shallow-disturbance increasing from 1% to 68% of plot area, but only covered 37% on sites of deep-disturbance. Within 2 years of harvesting, vegetation recovered to 80% cover on sites of shallow disturbance and 30% cover on deeply disturbed sites. Comparisons of sediment generated by sheetwash erosion post-harvesting with that derived by slope scalping during harvesting and, periodically by storm-induced landsliding during the post- harvest period, provided a process-based understanding of the relative importance of these sources for Compartment 49 (Table 1). Sediment generated from deeply disturbed sites occurred in two phases-initially by ‘scalping’ and then by surface wash. A single heavy rainfall event in April 2001, with an estimated return period of 2 years, initiated shallow landslides and accounted for the greatest proportion of the sediment total derived from the logged setting.

Table 1 Process-based rates of sediment generation and surface lowering following harvesting in Compartment 49 at Whangapoua Forest, Coromandel. Sediment Area Total Sediment Surface generating (ha) sediment generation lowering area (t) (t/ha) (mm) Undisturbed 14.5 (40%) 0 Na Na Roads & 2 (6%) Na Na Na landings Shallow 16 (43%) 16 1 0.08 Disturbance Deep 3.6 (10%) 57 16 1.3 Disturbance Landslides 0.4 (1%) 600 1500 125 n=36 Scalped areas 3.6 (10%) 1200 333 28 TOTAL 36 (100%) 1873 Mean value 51 4.2

Figures in brackets are the % of total area Na – not assessed. Note: Scalped area is classed as deeply disturbed

The rainfall event in April 2001 that initiated shallow landslides not only accounted for the greatest proportion of the sediment total derived from the logged setting but also that delivered to stream channels during the study period to date. The 9 landslides that reached a permanent stream channel collectively delivered 72% of the sediment total to streams. Areas that were ‘scalped’ as a consequence of the harvest operations, and directly connected to the streams, delivered 26% of the total sediment. By contrast, rain-drop and sheetwash processes on these same sites delivered less than 2% of the sediment total delivered to streams during the 16-month post-harvest period (Table 2).

Table 2 Sediment generation and delivery to streams Sediment Area connected to Sediment Sediment delivered to generating area stream generated stream (ha) (t) (t)

Roads & Landings n/a n/a n/a Undisturbed n/a n/a n/a Shallow disturbace n/a n/a n/a Deep Disturbance 0.18 2.9 2.9 (1.3) Scalped sites 0.18 60.0 60.0 (26.3) Landslides 0.07 (n=9) 330.0 165.0*(72.4) Totals 0.40 392.9 228.0 (100) N/a not assessed Figures in parentheses are % of total sediment delivered to permanent stream channels * 50% of sediment generated by 9 landslides remained on slope as debris tail and levee deposit

For environments such as the Coromandel, there is little that can be done to reduce the incidence of operational scalping. Without excessively high towers on haulers, the terrain makes it difficult to get enough lift to keep logs clear of the ground. In general, there is a tradeoff between accepting a certain amount of soil disturbance and having to add extra roads and landings. A number of studies from both New Zealand and overseas that have found that roads and landings are primary sources of sediment and any increases in the number of roads or landings would tend to increase the overall sediment being generated (Fransen et al. 2001).

Many of the effects of harvesting are relatively short-lived. Once vegetation recovers, many of the bare areas are no longer visible and, if they had been contributors of sediment early on, they would no longer be supplying sediment.

Conclusions Comparisons of rates of surface lowering following harvesting indicate that on weathered andesitic terrain in the Coromandel area soil depletion from sites of shallow-disturbance is an order of magnitude less than for deep-disturbance sites. In addition, rates of surface lowering determined for deep-disturbance sites are 3 times greater than for similar sites on pumiceous terrain and, between 6-10 times less than that on fractured sedimentary sandstone/mudstone lithologies (Marden & Rowan 1997; in prep). Across all terrains and lithologies, the rate of surface lowering on sites of deep disturbance on forest cutover is likely to be between one and three orders of magnitude less than that for compacted tracks, unsealed roads and haul paths.

Landslides generated 1500 t of sediment per hectare while areas scalped during the harvest operation generated 333 t of sediment per hectare resulting in rates of surface lowering of 125mm and 27.8mm, respectively. Sediment generation rate from all sources was 51 t/ha, a surface lowering rate across the catchment of 4.2mm. Had harvesting coincided with a storm-free and landslide-free period the rate of surface lowering would have been 0.8mm.

Finally, the take home messages are: 1. not all bare ground is bad, nor is it all eroding and contributing sediment; 2. landslides are the biggest contributors of sediment and they are difficult to manage for both in time and in space; 3. operational “erosion” such as scalping needs to be traded off against increased roads and landings in steep terrain.

References Fransen, P.J.B.; Phillips, C.J.; Fahey, B.D. 2001: Forest road erosion in New Zealand: overview. Earth Surface Processes and Landforms 26: 165-174.

Marden, M.; Rowan, D. 1995: Assessment of storm damage to Whangapoua Forest and its immediate environs following the storm of March 1995. Landcare Research Contract Report LC9495/172 for Ernslaw One Ltd.

Marden, M.; Rowan, D. 1995: Relationship between storm damage and forest management practices, Whangapoua Forest. Landcare Research Contract Report LC9495/173 for Ernslaw One Ltd. New Zealand Meterological service 1980: Depth-Duration-Frequency Tables based on daily rainfalls. New Zealand Meterological Service miscellaneous publication 162: (Supplement [1]). Phillips, C.J.; Marden, M. 1999. Review of vegetation-slope stability in plantation forests and risk assessment of Ohui Forest to landsliding. Landcare Research Contract Report for Carter holt Harvey Forests Ltd. LC9899/66.

Quinn, J.M.; Stroud, M.; Parkyn, S.; Burrell, G. 1995: Impacts of an intense rainstorm on 3-4 March 1995 on Whangapoua Forest landscape and streams. NIWA contract report (unpubl.) 14 p.

Williams, M. 1995: Whitianga Storm Event - 4th March 1995. Environment Waikato memorandum file 410003 (unpubl). 4p.