pre- and post-reforestation gully development in mangatu ... et al.pdf · pre- and...
TRANSCRIPT
RIVER RESEARCH AND APPLICATIONS
River Res. Applic. 21: 757–771 (2005)
Published online in Wiley InterScience(www.interscience.wiley.com). DOI: 10.1002/rra.882
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT IN MANGATUFOREST, EAST COAST, NORTH ISLAND, NEW ZEALAND
MICHAEL MARDEN,a* GREG ARNOLD,b BASIL GOMEZc and DONNA ROWANa
aLandcare Research Ltd, PO Box 445, Gisborne, New ZealandbMassey University, Palmerston North, New Zealand
cGeomorphology Laboratory, Indiana State University, Terre Haute, IN 47809, USA
ABSTRACT
Following clearance of the indigenous forest and conversion of the land to pasture early in the 20th century, gully erosionbecame a pervasive feature in the headwaters of the Waipaoa River basin, and was notably problematic in the 140-km2 areanow covered by the Mangatu Forest. In this area, before reforestation in 1961, gully erosion affected c. 4% of the terrain. After a24-year exotic reforestation programme the area affected by gullies was reduced to 1.5%, but of the eight gullies larger than10 ha in 1960 none had stabilized by 1988, although four had at least halved in size. Estimates that a gully will stabilize orincrease in size under a range of conditions suggest that in the case of gullies <1 ha in area, formed in terrain underlain byCretaceous rock, there is a >80% probability of stabilization after one forest rotation (c. 24 years). For gullies between1 and 5 ha in area the probability of stabilization is c. 60%. Gullies of 5 ha have an even chance of stabilizing over the timeframe of a single rotation. The key determinant is gully size and shape at the time of planting and, within this size range, theserelationships were stronger for linear than for amphitheatre-shaped gullies. Between 1939 and 1988 sediment production fromgullies in the portion (76%) of the Mangatu Forest underlain by Cretaceous-aged rock was c. 22 000 t km�2 yr�1, and during theperiod of maximum sediment production (1939–1960) they may have accounted for c. 17% of the Waipaoa River’s averageannual suspended sediment load. Reforestation reduced the contribution to c. 8% in the period between 1970 and 1988.However, the off-site (downstream) impact of sediment generated by the remaining 420 active gullies in the Waipaoa Rivercatchment is significant, not least on the capacity of the scheme that protects high-value agricultural land on the PovertyBay Flats from flooding. A targeted reforestation programme may be an alternative to raising the height of the existingartificial levees. It is estimated that additional exotic plantings totalling c. 15 400 ha (c. 7% of the Waipaoa River basin area)would produce a>64% reduction in sediment production from gullies on pastoral hillslopes within one forest rotation (c. 24 yr).Copyright # 2005 John Wiley & Sons, Ltd.
key words: gully erosion; reforestation; gully stabilization; sediment production; Mangatu Forest
INTRODUCTION
Erosion . . . (predominantly gullies, earthflows and shallow landslides) . . . is as spectacular in its quick and oftenirreversible degradation of otherwise good land as it is tragic in its economic implications for individual prop-
erty owners and for the district as a whole. Where such erosion has developed on a large scale the cost of erosion
control far exceeds the value of pastoral production from the land concerned. There is therefore no known way
of economically controlling erosion of this kind other than by complete reforestation. (NWASCO, 1970)
Accelerated erosion by gullying can make a major contribution to the sediment yield of steepland catchments (Le
Bourdiec, 1972; De Rose et al., 1998). Gully initiation is commonly explained in terms of process thresholds for
incision (Patton and Schumm, 1975; Prosser and Abernethy, 1996), and is often linked with the degradation or
eradication of vegetation cover (Lyell, 1849; Graf, 1979; Prosser and Slade, 1994; Prosser and Soufi, 1998). Stu-
dies of the morphology and evolutionary development of gullies (relatively deep and rapidly eroding channels)
Received 8 February 2004
Revised 6 August 2004
Copyright # 2005 John Wiley & Sons, Ltd. Accepted 12 August 2004
*Correspondence to: Michael Marden, Landcare Research Ltd, PO Box 445, Gisborne, New Zealand.E-mail: [email protected]
suggest they have a limited lifespan and rapidly evolve to a condition of relative stability (Ireland et al., 1939).
Some gullies appear to be a natural component of landscape evolution, but many contemporary gullies formed
after native forests were cleared and agriculture intensified in the 19th and 20th centuries (Ireland et al., 1939;
Wells and Andriamihaja, 1993; Prosser et al., 1994; Harvey, 1996; Belyaev et al., 2004). In either case, gully exten-
sion represents a major adjustment to the landscape that is imprinted on the drainage network and has a profound
impact on basin sediment yield.
Concern over accelerated erosion induced by human activity has generated a substantial literature on gully
development (Harvey et al., 1985), not least in New Zealand where, in the North Island, gully erosion affects some
10% of the land area (Eyles, 1985). However, it remains that in locales where gully erosion has developed on a
large scale the cost of erosion control often far exceeds the value of production from the land concerned
(NWASCO, 1970), and the most economic way of controlling erosion of this kind is by reforestation (see Piegay
et al., 2004).
Government-funded reforestation programmes over the past 40 years have had considerable on-site success in
reducing gully erosion in the headwaters of the Waipaoa river basin, North Island, New Zealand (Figure 1), which
was initiated following clearance of the indigenous forest by European settlers. However, the literature describing
the ameliorating influence of the exotic forests is, in large part, anecdotal (Allsop, 1973), seldom quantitative
(Hicks, 1991; Phillips et al., 2000; Gomez et al., 2003), and has never been fully evaluated. In this paper we
quantify changes in gully area over a 21-yr period of pastoral production followed by a 24-yr reforestation
period, and evaluate the contribution sediment derived from the gully-prone Cretaceous-age lithologies makes
to the Waipaoa River system, for the periods 1939–1960, 1960–1970, 1970–1988, using the relationship derived
by De Rose et al. (1998). We also discuss the long-term implications of sediment production from treated and
untreated gullies elsewhere in Waipaoa basin, in the context of the capacity of the Waipaoa River Flood Control
Scheme. Our results are relevant to other areas of New Zealand where gully erosion is still widespread, and refor-
estation is, as it has been in the past, the preferred management option for erosion control on 195.5 km2 of land
identified as severely eroding soft rock terrain in the East Coast region (Parliamentary Commissioner for the
Environment, 1993).
STUDYAREA
The 140-km2 Mangatu Forest is located in the headwaters of the 2200-km2 Waipaoa River basin (Figure 1).
Climate in the study area is temperate maritime, with warm, moist summers and cool, wet winters. At the lowest
elevation (200m) annual precipitation is 1350mm, increasing to 2500mm at elevations of 800m (Pearce et al.,
1987). Tropical cyclones periodically accelerate both gully and earth flow activity. The largest recorded cyclonic
storm (Bola) occurred in March 1988, and generated 500 to 700mm of rain in a 5-day period. Since the turn of the
20th century there have been 29 extreme rainfall events, when the discharge of the Waipaoa River (at Kanakanaia)
exceeded 1500m3 s�1, and there is evidence of widespread shallow landsliding and accelerated gully erosion
(Cowie, 1957; Phillips et al., 1990). There is a 29% chance that an extreme rainfall event will occur every year,
and a >99% chance that such an event will occur every decade (Kelliher et al., 1995).
Clearance of indigenous podocarp-hardwood forest began c. 1894, and had essentially been completed by 1914
(Black, 1977). Photographs (c. 1903–1910) suggest the terrain revealed by the clearances was susceptible to mass
movements (that is, large, deep-seated earthflows and shallow landslides). Mass wasting accelerated and gullies
were initiated following the change in soil moisture status, pattern of hillslope runoff (from subsurface, diffuse
drainage, to surface runoff and its concentration along preferred drainage channels), and the loss of root strength
after deforestation (O’Loughlin, 1974a, 1974b). The chronology of the early stages of gully development is anec-
dotal, but it has been suggested the acceleration of mass wasting and gully initiation in the upper Waipaoa basin
took place in the first decade of the 20th century (Hill, 1895; Henderson and Ongley, 1920; Hamilton and Kelman,
1952; Gage and Black, 1979). The largest (Tarndale) gully complex (Figures 1 and 2) is thought to have been
initiated in the winter of 1915 on the site of an extant slump, and many other gullies were initiated during the
subsequent winters of 1916, 1917 and 1918. By 1910–1912, gully-derived sediment had begun to impact on
the river system, and in headwater streams the cobble-sized bed material had been replaced by fine gravel and sand
758 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
Figure 1. Location of Mangatu Forest area affected by gullies during the period 1939–1988 as interpreted from aerial photography(‘T’ indicates location of the Tarndale gully complex), and geological formations (after Mazengarb and Speden, 2000)
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 759
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
Figure 2. Pre- (1961) and post-reforestation (1988) photography of a large gully-mass movement complex known locally as Tarndale Slip.When planted (1964) with Douglas fir (Pseudotsuga menziesii) this gully was 21.4 ha in size. Early (1950s) attempts to retain sediment withinthe upper reaches involved the construction of fascines using brushwood supported by live plantings of Populus and Salix species (inset).Inevitably such structures within large gully systems were overwhelmed (circled) by sediment. Similar structures within smaller gullies(<1 ha) often proved successful. While reforestation of this gully has not significantly reduced sediment production from it, by 2004 the40-yr-old plantings had nonetheless succeeded in stabilizing slopes downstream of the gully-head to reduce the overall area of active gullyerosion to 13.8 ha. The elevation difference between stream level and the ridge at the head of Tarndale Slip is 220m. (1961 photograph courtesyof J. Johns; inset courtesy of Ministry of Works, NZ; 1961 photograph and inset reproduced by permission of New Zealand Forest Research
Institute Limited. 1988 photograph reproduced by permission of N. Trustrum, Geological and Nuclear Sciences)
760 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
(Jones and Howie, 1970; Black, 1977), and by the late 1920s channel aggradation and widening was a noticeable
and widespread phenomenon (Hamilton and Kelman, 1952; Gage and Black, 1979).
Gullies in the Mangatu Forest are associated with three lithological assemblages (Mazengarb and Speden, 2000;
Figure 1). The Whangai Formation (Wh) comprises variably indurated, well-bedded, alternating siliceous argillite,
mudstone and sandstone together with poorly bedded, calcareous mudstone of Late Cretaceous to Palaeocene age.
The Early to Late Cretaceous Tikihore Formation (Tk) consists of metre-bedded, alternating sandstone and mud-
stone displaying evidence of localized folding and soft sediment deformation. Both formations are part of the East
Coast Allochthon, and are thrust against the less deformed, sandstone and mudstone of the Early to Middle
Miocene age Tolaga Group (TG). Hereafter we use the terms Cretaceous terrain (76% of the study area, and synon-
ymous with the Whangai and Tikihore Formations) and Tertiary terrain (24%, synonymous with the Tolaga
Group). Within the Mangatu Forest area contacts between all three lithological assemblages are generally faulted,
and hillslopes are mantled by friable tephric coverbeds (Gage and Black, 1979; Black, 1980).
Gully distribution and type are strongly influenced by slope physiography, which in turn reflects the underlying
geology. Major faults, joint patterns and fold structures predispose rocks in the study area to mechanical disinte-
gration under the influence of water, and may have influenced the development and orientation of many gullies
(Gage and Black, 1979; Black, 1980). Several of the largest gully complexes (e.g. the Tarndale and Mangatu gully
complexes) occur in association with zones of extensive crushing along major faults (Black, 1980; Mazengarb
et al., 1991) and with argillite rocks that are especially susceptible to acid sulphate weathering (Pearce et al.,
1981). In these amphitheatre-shaped gullies, fluvial erosion occurs in association with rotational slumping (see
Betts et al., 2003), as the gully inexorably cuts back into the disturbed terrain (Pearce et al., 1987; De Rose
et al., 1998). Linear-shaped gullies are often associated with earthflow topography, where the long (c. 800m)
20� to 35� slopes have re-graded to the modern-day river level in response to a period of rapid, postglacial down-
cutting (Berryman et al., 2000; Eden et al., 2001). Many such gullies are therefore found on slopes immediately
adjacent to the main stem of the Waipaoa River, where incision was facilitated by this river’s alignment along a
shear zone within Whangai Formation. In contrast, the extensive areas of earthflow topography associated with
Tikihore Formation have been stable for c. 10 000 years (Black, 1980). Here, slope adjustment in response to post-
glacial downcutting is incomplete and the gullies are less well developed.
Initial attempts to control gully erosion with fascines and check dams were, in the long term, largely ineffective.
Many such structures were overwhelmed within a few years of establishment by sediment emanating from gullies
(Figure 2). Increasing costs associated with on-farm gully stabilization led the New Zealand government to the
purchase of large tracts of farmland in the headwaters of the Waipaoa River basin for reforestation. Reforestation
began in 1960, and by 1987 c. 140 km2 of exotic forest been established (Figure 3). The principal tree species were
Pinus radiata (69% of forested area), and Pseudotsuga menziesii (8.5%). Minor species include Pinus nigra, euca-
lyptus species and Acacia melanoxylon (3.5%). Unplanted areas within the study area including indigenous rever-
sion account for 1.5%, pasture 14% and aggraded riverbed 3.5%. Poplar and Salix species were also planted within
gully systems to help stabilize sideslopes and reduce channel scour (Dolman, 1982). Planting was undertaken in
two periods, 1961 to 1971 and 1972 to 1987 (Figure 3). The first areas to be planted were stocked with 2200 to
1500 stems per hectare, but the planting density was reduced to between 1500 and 1250 stems per hectare during
the latter period (the piecemeal approach to reforestation prevents us from differentiating between effects due to
different tree species, planting densities, and tending regimes).
METHODS
Data sources
The planimetric area of individual gullies was measured from aerial photographs acquired in 1939, 1960, 1972
and 1988 (Figure 3). The earliest photography was acquired c. 30 yr after the indigenous forest had been removed,
the 1960 coverage coincided with the beginning of reforestation, and the latest coverage was flown at the end of the
replanting programme, after the Cyclone Bola storm impacted the study area. Gully outlines were traced and trans-
ferred to a 1:6700-scale contoured transparency. After the outlines had been digitized, the area of each gully was
determined, and the difference in area between successive time periods computed. Date(s) of planting, area and
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 761
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
species planted (as noted in compartment records) were also incorporated in the database, as was the lithology of
the terrain that supported each gully. Changes in total number, total area and an indication of when gullies first
appeared, for the period 1939 to 1988, are presented in Figure 4.
Derived variables and statistical analysis
The shape of each gully was defined as 2�� area/(perimeter)2 (circular gullies approach the maximum value of
0.5 and values for elongate gullies tend to zero), and the (maximum) area affected by gully erosion was computed
as the aggregate change in gully area recorded between 1939 and 1988 (Figure 1).
Because there were many small and few large gullies, the distribution of gully area was highly skewed. Con-
sequently, log(area) was used in all analyses to discriminate between the smaller-sized gullies (see Figure 5). This
suggests that ‘area effects’ are important and that they relate to proportional rather than absolute change.
Analysis looked for relationships between the proportional change in area or the probability of a gully stabiliz-
ing during a given period as response variables, and the area, slope, shape, altitude and years since planting as
explanatory variables. Splus (2002) was used for all analysis.
Classification trees (Clark and Pregibon, 1992), explore these relationships without making assumptions
about their form. These find yes/no decision rules based on gully properties that best predict whether a gully will
Figure 3. Distribution of gullies considered to be actively producing sediment: 1939, 1960, 1970 and 1988
762 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
stabilize. Logistic regression then developed the relationships that classification analysis found important into a
model that estimated the probability of a gully stabilizing under a range of conditions, including the time since
planting. Incorporating spatial correlation did not improve the fit and gave virtually the same estimates.
Sediment production
De Rose et al. (1998) used elevation differences between high-resolution digital elevation models constructed
from sequential aerial photography as a basis for estimating the volume of sediment removed from a small sample
(11) of the gullies in the Mangatu Forest. All the gullies lie within Cretaceous terrain, and 10 drain to Te Weraroa
Stream (Figure 1). The range of gully sizes is similar to that in the Mangatu Forest as a whole, and Betts and De
Rose (1999) developed a regression equation (r2¼ 0.996) that is applicable to both linear and amphitheatre-shaped
gullies at all stages of development:
y ¼ 460þ 2750xþ 160x2
where y is total sediment production from a gully (m3 yr�1) and x is gully area (ha). This equation was used to
estimate sediment production from gullies developed in Cretaceous terrain in the study area as a whole (see Gomez
et al., 2003). Betts and De Rose (1999) established that there is a different relationship for gullies in Tertiary ter-
rain. However, few large gullies (>5 ha in area) were included in the sample used to derive the relationship; there-
fore, the sample is not likely to be representative of the Tertiary terrain in the Mangatu Forest as a whole.
Figure 4. Change in total number (A) and total area (B) of all gullies for the period 1939 to 1988. The different shadings indicate when a gullyfirst appeared
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 763
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
Consequently, the sediment budget in this paper applies to three time periods (1939–1960, 1960–1970 and 1970–
1988) and to gullies within Cretaceous terrain. The mass of material supplied by these gullies during each time
period was computed using a dry bulk density of 2000 kgm�3 (Phillips, 1988). Gauging data for the periods
1960–70 and 1970–88 were used (D. M. Hicks pers. comm.) to estimate the proportional (%) equivalent sediment
from gullies in the Cretaceous terrain made to the average annual suspended sediment yield of the Waipaoa River.
Suspended sediment data were not collected before 1960, and the average annual suspended sediment yield for the
period 1939–1960 was estimated on the basis of the proportional difference in the annual rate of sediment produc-
tion from gullies, relative to that produced by gullies in the period 1970–1988 (the timespan for which the data are
considered most reliable). The estimates represent an upper bound because no account is taken of deposition in
feeder channels and on fans, or of the proportion of coarse sediment (gravel) generated by gully erosion (Gomez
et al., 2003).
RESPONSE TO REFORESTATION
Historical analysis of gully stabilization
The composite area affected by gully erosion between 1939 and 1988 was 895 ha (c. 6.4% of the study area;
Figure 1). Once it had peaked in the early 1960s, during the 24-yr planting programme (1961–1985; <1% of the
forest was replanted after this date) the total area affected by gully erosion decreased by 64% (most dramatically in
those areas planted between 1961 and 1970 in the Cretaceous terrain). From the earliest available photography,
Figure 5. Changes in the distribution of gully area (A) and shape (B) for the periods 1939, 1960, 1970 and 1988. The dot is the median, the boxindicates the range of the middle half and the brackets indicate the extent of the distribution. The circles highlight gullies larger than reasonablebounds for this distribution, indicating amphitheatre-shaped gullies >10 ha. The shape index ranges from 0.5 for a circle (amphitheatre-shaped
gully) to 0 for a line (linear gully)
764 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
taken three or four decades after deforestation, gullies in the Mangatu Forest ranged in size from 0.02 to 26 ha
(Figure 5(A)), and occupied a total area of 305 ha (c. 2% of the study area; Figure 4(B)). Between 1939 and
1960 median gully size increased fivefold (Figure 5(A)). Most gullies were linear in shape (Figure 5(B)), and some
had stabilized in response to the encroachment of bracken and native regenerating species (Allsop, 1973). In the
case of gullies<0.5 ha in area in 1939, 65% had stabilized by 1960, and 35% had increased in size; 62% of gullies
0.5 to 1 ha in area increased in size; as did 80% of gullies 1 to 5 ha in area. Total gully area initially increased
at a rate of c. 2.5% per year (c. 8 ha yr�1), but during the period of maximum gully expansion (1939–1960)
this increased to 14 ha yr�1, and by 1960 the gullied area occupied (a maximum of) c. 4% of the study area
(Figure 4(B)).
By 1970, 45% of the Mangatu Forest had been replanted (Figure 3) and the combined area occupied by active
gullies had declined to c. 2.6% (Figure 4(B)). Within the planted area, median gully size had decreased to 0.4 ha
(Figure 5(A)), and 85% of the gullies had decreased in size. Only 4% of the active gullies were larger than 5.6 ha in
area, and 58% were >0.5 ha. Thus within the initial 9-yr period following planting, the gullied area decreased by
39% (see Figure 3, 1960 and 1970). Beyond the reforested area, however, new linear gullies continued to develop
(Figure 4(A) and 4(B)), and existing gullies continued to increase in size (Figure 5(A)).
The remaining area within the forest boundary was, in large part, replanted between 1970 and 1985. During the
24-yr replanting period median gully area increased to 0.8 ha (Figure 5(A)); reforestation had stabilized many of
the smaller gullies and only medium- to large-sized gullies remaining active. By 1988, reforestation had success-
fully reduced the area of the Mangatu Forest impacted by active gullies to just 1.5% (216 ha). Even a large cyclonic
storm (Cyclone Bola) in March 1988 triggered the collapse of only one, partially forested (4.3 ha) gully, and <1%
of the forested area within the study area showed signs of renewed slope instability. Of the gullies that have
remained active, most are located in steeper terrain and are of the larger, amphitheatre-shaped type (see De Rose
et al., 1998).
Prediction of gully stabilization
The logistic model which has been developed to predict gully stabilization is relevant to gully-prone hill country
in areas of Cretaceous terrain where reforestation is likely to be considered as an erosion-control option. Probabil-
ities of gullies 0.5 ha, 1 ha and 5 ha stabilizing following a range of reforestation periods are presented (Figure 6).
The probabilities estimated by this model were based on the proportion of gullies in 1960 that had zero area in
1988. Details of the fitted model are as follows:
Response: Log odds of a 1960 gully in Cretaceous terrain stabilizing by 1988.
Variables in explanatory function: Constantþ log (1960 area)þ 1960 shapeþ 1960 altitudeþ years since
planting.
Figure 6. Predicted probability of gully-stabilization plotted against years since planting for a 0.5-ha, 1-ha and 5-ha sized gully associated withCretaceous terrain
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 765
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
Dispersion parameter for binomial family¼ 1.
Null deviance: 261.1 on 255 degrees of freedom.
Residual deviance: 200.9 on 251 degrees of freedom.
Area at the time of planting was the key determinant of the probability that reforestation would stabilize a gully
and so reduce sediment production, with a >80% probability that a gully less than 0.5 ha would stabilize after one
rotation (Figure 6) and a 60% probability of gullies between 1 and 5 ha stabilizing. Gullies of 5 ha have an even
chance of stabilizing over the time frame of a single rotation, but of the eight gullies larger than 10 ha in 1960 none
had stabilized by 1988, although four had at least halved in size. The reduction in gully planimetric area is attrib-
uted to the influence of the newly established forest and to the effectiveness of the forest cover in influencing the
hydrology of and erosion activity within the gullies. A mature, closed-canopy stand of Pinus radiata has the poten-
tial to reduce runoff by between 25% and 30% (Pearce et al., 1987), and tree roots also improve gully-side stability.
The critical size for reforestation appears to be of the order of 0.5 to 1 ha, and the relationship between forest-
planting date and gully planimetric area suggests gully size decreases with increasing tree age. In linear gullies
it is often possible to plant to the channel edge and thus the forest canopy may span the entire gully to provide
effective stability sooner. Amphitheatre-shaped gullies are more difficult to stabilize because gully width typically
exceeds gully length and replanting creates insufficient canopy cover to effect a reduction in surface runoff and
mitigate erosion. The model parameters quantified the difference, estimating that the odds of a gully with a shape
index of 0.1 (i.e. quite linear) stabilizing are seven times those for a gully with shape index 0.4 (i.e. quite round).
Reforestation has proved to be largely ineffective in ameliorating erosion in gully complexes >20 ha in area,
though in most instances partial stabilization was evident (Figure 2). The survival of forest plantings within
once-active gullies, channel incision into depositional fans emanating from them and the subsequent stabilization
of these fan deposits (Figure 7) is further evidence that reforestation has been effective in stabilizing all but the
largest and most active of gullies. Despite a major cyclonic event in 1988 (Cyclone Bola) all but one gully
remained stable.
Sediment production from gullies
Within Mangatu Forest, gully density was always highest within the Cretaceous terrain (5 per km2) and lowest
(2 per km2) within the Tertiary terrain. In the period 1939–1970, c. 78% of the total gullied area occurred in the
Cretaceous terrain (107 km2), which is predisposed to the development of large gully complexes (Pearce et al.,
1981). Sediment production from gullies within the Cretaceous terrain during the 1939–1960 pre-reforestation
period was c. 27 000 t km�2 per year, increasing to c. 30 000 t km�2 per year during the 1960–1970 period, before
decreasing to c. 11 000 t km�2 per year during the 1970–1988 period, by which time most of the reforested area
had reached maturity. Total sediment production from gullies in Cretaceous terrain was 113Mt for the 1939–1988
period.
In the pre-reforestation period (1939–1960), gullies within Cretaceous terrain probably generated the equivalent
of c. 17% of the Waipaoa River’s average annual suspended sediment yield (Table I). The total contribution from
all gullies within the Mangatu Forest at this time, and elsewhere in the Waipaoa River basin, is uncertain. However,
the 1939 aerial photography shows that gully erosion was widespread in the Tertiary terrain within the Mangatu
Forest (see Figures 1 and 3), and also throughout the Waingaromia and Waihora sub-catchments (Figure 1). It is
estimated that gullies within the latter two sub-catchments generated c. 33% of the Waipaoa River’s suspended
Coefficients of variables:
Variable Value Standard error t ratio
Constant 15.4 2.4 6.34
Log (1960 area) �1.35 0.21 �6.31
1960 shape �6.68 2.34 �2.86
1960 altitude �0.003 93 0.001 65 �2.38
Years since planting 0.0777 0.0363 2.14
766 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
yield (Griffiths, 1982), a contribution that was probably sustained until they were reforestated in the late 1980s.
Thus, taking into account the potential sediment contribution generated from all gullies within Waipaoa River
basin, gully erosion is likely to have contributed >50% of the Waipaoa River’s suspended sediment load in the
pre-reforestation period.
The contribution of sediment from the four largest gully complexes in the periods 1939–1960, 1960–1970 and
1970–1988 was 11%, 28%, and 35%, respectively. In 1988 their collective area represented 33% of the total gully
area, and in the period 1939–1988 they produced 19% of all sediment generated. Subsequent to reforestation their
relative contribution increased significantly. This reflects the stabilizing and consequent reduction in sediment
production from the more numerous, smaller gullies as the exotic forest matured (Gomez et al., 2003). The
remaining gully complexes are thus likely to dominate sediment production from the Mangatu Forest for
decades into the future, even though overall sediment production since planting began has declined by c.
40Mt (Table I).
The largest and best known of the gully complexes within the Mangatu Forest, the Tarndale Gully complex
(Figures 1 and 2), was a landmark feature in 1915 (Allsop, 1973) that produced c. 9% of the total sediment
(113Mt) generated from all gullies during the study period as a whole (1939–1988). Before reforestation this gully
complex is estimated to have produced c. 5% of the sediment load generated by all gullies within the Cretaceous
Table I. Sediment production from gullies in Cretaceous terrain and their contribution to the average annual suspended sedi-ment yield of the Waipaoa River
1939–60 1960–70 1970–88
Total sediment production from gullies (Mt) 60.84 31.68 20.82Annual sediment production from gullies (Mt yr�1) 2.9 3.2 1.2Average annual suspended sediment yield of Waipaoa River (Mt yr�1) 16.7* 17.0y 15.0yPercentage contribution to suspended sediment yield 17 19 8
*Estimated from the proportional difference in the annual rate of sediment production from gullies relative to that for the 1970–1988 period.yD. M. Hicks (pers. comm.).
Figure 7. Pre- (1961) and post-reforestation (1972, 2004) photography of a medium-sized gully in Te Weraroa Stream, Mangatu Forest. Pinusnigrawas planted (1962) on the lesser-eroded interfluves surrounding this gully; then in 1966 Pinus radiatawas planted on the steeper and moreseverely eroding slopes immediately flanking the gully and within the gully itself. By 1972 the effectiveness of the plantings in stabilizing thegully were apparent and its channel incised below the level of the fan at the mouth of the gully. Further within-gully plantings (1974) ofP. radiatawere undertaken on the remainder of bare slopes and of the fan. Before planting, this gully was 7.6 ha in size, and by 1988 reforestation hadreduced the area of active erosion to 0.8 ha. The latest photograph shows that in spite of a major cyclonic event in 1988 (Cyclone Bola) this andsimilarly reforested gullies of this size, have remained stable. The elevation difference between stream level and the ridge at the head of thisgully is 280m. (1961 and 1972 photographs courtesy of J. Johns and reproduced by permission of New Zealand Forest Research Institute Lim-
ited; 2004 photograph was taken by R. Hambling and reproduced by permission of R. Hambling, Ministry of Agriculture and Forestry)
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 767
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
terrain. As more of the Mangatu Forest was reforested, the relative contribution from Tarndale Gully complex
increased, and in the 1970–1988 period reached c. 14% (see Gomez et al., 2003).
IMPLICATIONS FOR FUTURE LAND AND RIVER MANAGEMENT
An initial response to the decrease in sediment production following reforestation has been channel narrowing and
incision. A similar response to reforestation has been widely documented, notably in Europe where major refor-
estation occurred a century ago (Piegay and Salvador, 1997; Garcia-Ruiz et al., 1997; Liebault and Piegay, 2001;
Surian and Rinaldi, 2003). In the Waipaoa catchment the reversal from channel aggradation to degradation first
became evident in zero-order basins (gullies) within a decade of reforestation (Figure 7). Here, channel incision
and narrowing has led to the formation of abandoned alluvial fans and degradational terraces that have since
become stabilized by a mix of invasive, indigenous and exotic weed, grass and shrub species. Higher-order chan-
nels are at a different stage in the aggradation/degradation cycle as the few remaining active gullies produce suffi-
cient sediment at the upstream end of the catchment to at least retain a steady state within these larger channels.
With little prospect of further reductions in sediment production from these gullies and in the absence of major
recurrent floods, it is considered unlikely that the Waipaoa riverbed, within the forested reach, will reflect the
effects of reforestation and adjust to net degradation for several decades.
The intensively cultivated floodplain along the lower reaches of the Waipaoa River, downstream from Kanaka-
naia (Figure 1), is protected from flooding by artificial levees (locally termed ‘stopbanks’), constructed since 1950
as part of the Waipaoa River flood control scheme. The scheme currently provides a notional ‘1 in 100 yr’ level of
protection against flooding, but is gradually losing capacity owing to aggradation of the riverbed and floodplain
(Gomez et al., 1998, 1999). Aggradation of the berms due to suspended sediment deposition during overbank
events is currently of much greater significance than channel aggradation. Bed load amounts to<1% of the annual
long-term suspended sediment load (15Mt) of theWaipaoa River, or c. 75 000m3 yr�1 (Trafford, 1998; Hicks et al.,
2000), and rates of channel aggradation at the upper end of the flood control scheme are of the order of 20mmyr�1,
compared with c. 40mmyr�1 on the floodplain (Gomez et al., 1999). Flows associated with overbank events trans-
port 24% of the mean annual suspended sediment load. It is anticipated that the height of the existing stopbanks
will have to be increased in coming decades. However, engineered solutions to the off-site impacts of land-use
change are becoming progressively more expensive and increasingly burdensome to ratepayers. Reforestation
has had a positive effect on the supply of coarse sediment to and rates of aggradation along streams in the basin
headwaters (see Gomez et al., 2003). For example, in the c. 3-km long reach immediately downstream of the con-
fluence with Te Weraroa Stream, the rate of aggradation in the Waipaoa River decreased by c. 50%, from 0.1–
0.14mmyr�1 in the 19 year period from 1947 to 1966, to 0.045–0.052mmyr�1 in the ensuing 34-yr period
(1966 to 2000). Viewed from this perspective, an alternative solution might be to limit the supply of (fine)
gully-derived sediment, a proportion of which is deposited on the floodplain during overbank events (see Gomez
et al., 1999), at source. Indeed, preliminary evaluations suggest targeted conservation (reforestation) projects could
provide substantial benefits to the flood control scheme by extending its design life in the medium to long term
(Peacock and Turner, 2003).
Today there are 420 currently active sediment-producing gullies within Waipaoa catchment, c. 25% of which
have been reforested. The 37 forested gullies within the Cretaceous terrain in the Mangatu Forest currently gen-
erate c. 8% of the Waipaoa River’s annual suspended sediment load, and unless remedial plantings are undertaken
within gullies that have remained active, sediment production is unlikely to decline further in future years. Without
treatment, the majority of the 289 gullies on pastoral hillslopes are also likely to expand in size, with a consequent
increase in sediment production, although it is expected that within the next 20 years the contribution of gully-
derived sediment from newly forested areas on Tertiary terrain elsewhere within the catchment will decline. For
gullies associated with areas of Cretaceous terrain there is a >80% probability that reforestation would ameliorate
sediment production from small gullies (<1 ha in area), and a 60% probability that it would impact on sediment
production from gullies between 1 and 5 ha in area. Experience also suggests that, in the case of c. 89% of the
untreated gullies within these size ranges on Tertiary terrain within this catchment, reforestation would be as
equally successful. Similarly, the size of gullies 5 to 10 ha in area (11% of all gullies) would be reduced by
768 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
c. 50%, but each of these gullies would probably remain a point source of sediment beyond the end of the first
forest rotation. Sediment production from large gullies (>20 ha in area), such as the Tarndale Gully complex,
appears to be largely unaffected by reforestation (see De Rose et al., 1998; Gomez et al., 2003), and these features
will probably continue to supply large amounts of fine sediment to the river system. De Rose et al. (1998) calcu-
lated gullies of this size generated the equivalent of between 2% and 3% of the total annual suspended sediment
load (10.7� 106 t a�1) at Kanakanaia. Using a revised annual suspended sediment load of 15� 106 t a�1 for the
1970–1988 period (Hicks, pers. comm.) and the estimated proportional (%) equivalents for the earlier periods
(Table I), Tarndale produced c. 1% of the river’s load during the pre-reforestation period (1939–1960), increasing
to c. 2% during the 1960–1970 period, and declining to c. 1% during the post-reforestation 1970–1988 period.
On the basis of the results of this study, it is reasonable to suggest that a targeted reforestation programme would
produce a>64% reduction in sediment production from the gullies on pastoral hillslopes within one forest rotation
(c. 24 yr). This could be effected by additional exotic plantings totalling c. 15 400 ha, which amounts to c. 7% of
the total Waipaoa River basin area (a smaller area would need to be planted if other treatment options (e.g. rever-
sion to native scrub, and poplar and willow plantings) were applied to the many small gullies). In the context of the
pastoral hillslopes on Tertiary terrain, the long-term benefits of a targeted reforestation programme with the objec-
tive of also stabilizing gullies, has to be viewed in concert with the commensurate off-site benefits that should
accrue from the associated reduction in the amount of sediment generated by shallow landslides (Reid and Page,
2002).
CONCLUSIONS
Gully erosion in the headwater reaches of the Waipaoa River basin was initiated in the first quarter of the 20th
century after the indigenous forest was cleared and the land converted to pasture. By 1939, three or four decades
after deforestation, gullies in the 140 km2 of the headwaters now covered by the Mangatu Forest ranged in size
from 0.02 to 26 ha, and occupied a total area of 305 ha (c. 2% of the study area). By the early 1960s the gullies
occupied a maximum of c. 4% of the study area, but 24 yr after reforestation with exotic species this had decreased
by c. 64%, to 1.5%. However, of the eight gullies larger than 10 ha in 1960 none had stabilized by 1988, although
four had at least halved in size. The key determinants affecting whether or not reforestation can facilitate stabiliza-
tion are the size and shape of each gully at the time of planting; linear gullies are more likely to become stable than
their amphitheatre-shaped counterparts. Probabilities that reforestation will be effective range from>80% for gul-
lies<1 ha in area to c. 60% for gullies between 1 and 5 ha in area. Gullies of 5 ha have an even chance of stabilizing
over the time frame of a single rotation. Erosion in the larger (>10 ha in area) gully complexes generally is too far
advanced to be mitigated by reforestation. The overall success of reforestation in ameliorating gully erosion in the
Waipaoa catchment can be attributed to: the selection of fast-growing tree species (which are harvestable c. 24 yr
after planting), ideal growing conditions, and the planting strategy adopted. That is, gully stabilization was
achieved first, by planting as much of the gully watershed area as physically possible and, second, by delaying
within-gully plantings until there was a noticeable reduction in runoff and sediment supply to the channel as evi-
denced by channel incision and fan abandonment. This usually coincided with canopy closure, c. 8 yr after plant-
ing. In the coming years, the large gully complexes will continue to dominate the sediment supply to the
headwaters of the Waipaoa River and the sediment generated by the 420 untreated and necessarily expanding gul-
lies in the headwaters of the Waipaoa River Basin could have a deleterious effect on the capacity of the scheme that
protects high-value agricultural land further downstream (on the Poverty Bay Flats) from flooding. However, the
requirement to upgrade the flood-control scheme by raising the height of the existing artificial levees (stopbanks)
could potentially be obviated by a targeted reforestation programme that would involve additional exotic plantings
totalling c. 15 400 ha. It is estimated this would produce a>64% reduction in sediment production from the gullies
on pastoral hillslopes within one forest rotation (c. 24 yr).
ACKNOWLEDGEMENTS
This work was supported by the Foundation for Research Science and Technology, contract number CO9X0013,
the Ministry of Agriculture and Forestry (East Coast Forestry Project, Gisborne), and National Science Foundation
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 769
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
grant BCS0136375 and BCS0317570 to BG. We thank the staff of Rayonier, East Coast, and Anne Sutherland of
Landcare Research Ltd, for contributing to the digitizing of maps. The manuscript benefited from comments
provided by Murray Hicks (NIWA), Les Basher and Mike Page (Landcare Research). Anne Austin edited the
typescript, and figures were drawn by Nicolette Faville.
REFERENCES
Allsop F. 1973. The Story of Mangatu. Government Printer: Wellington, New Zealand.
Belyaev VR, Wallbrink PJ, Golosov VN, Murray AS, Sidorchuk AY. 2004. Reconstructing the development of a gully in the upper Kalaus basin,
Stavropol region (southern Russia). Earth Surface Processes and Landforms 29: 323–341.
Berryman K, Marden M, Eden D, Mazengarb C, Ota Y, Moriya I. 2000. Tectonic and paleoclimatic significance of Quaternary river terraces of
the Waipaoa River, east coast, North Island, New Zealand. New Zealand Journal of Geology and Physics 43: 229–245.
Betts HD, De Rose RC. 1999. Digital elevation models as a tool for monitoring and measuring gully erosion. Journal of Applied Earth
Observation and Geoinformation 1: 91–101.
Betts HD, Trustrum NA, De Rose RC. 2003. Geomorphic changes in a complex gully system measured from sequential Digital Elevation
Models, and implications for management. Earth Surface Processes and Landforms 28: 1043–1058.
Black RD. 1977. Rivers of change; early history of the upper Waipaoa and Mangatu catchments. Unpublished Report, New Zealand Forest
Service, Forest Research Institute: Christchurch, New Zealand.
Black RD. 1980. Upper Cretaceous and Tertiary geology of Mangatu State Forest, Raukumara Peninsula, New Zealand.New Zealand Journal of
Geology and Geophysics 23: 293–312.
Clark LA, Pregibon D. 1992. Tree based models. In Statistical Models in S, Chambers, JM, Hastie TJ (eds). Wadsworth & Brooks/Cole: Pacific
Grove, CA; 377–417.
Cowie CA. 1957. Floods in New Zealand 1920–1953. The Soil Conservation and Rivers Control Council: Wellington, New Zealand.
De Rose RC, Gomez B, Marden M, Trustrum NA. 1998. Gully erosion in Mangatu Forest, New Zealand, estimated from digital elevation
models. Earth Surfaces Processes and Landforms 23: 1045–1053.
Dolman K. 1982. Gully revegetation on the East Coast—with a focus on Acacia. New Zealand Forest Service: Gisborne.
Eden DN, Palmer AS, Cronin SJ, Marden M, Berryman KR. 2001. Dating the culmination of river aggradation at the end of the last glaciation
using distal tephra compositions, eastern North Island, New Zealand. Geomorphology 38: 133–151.
Eyles GO. 1985. The New Zealand Land Resource Inventory Erosion Classification. Ministry of Works and Development, Soil Conservation
Centre, Aokautere, Water and Soil Miscellaneous Publication no. 85. Ministry of Works: Palmerston North, New Zealand.
Gage M, Black D. 1979. Slope-stability and geological investigations at Mangatu State Forest. New Zealand Forest Service Technical Paper
no. 66. New Zealand Forest Service: Wellington.
Garcia-Ruiz JM, White SM, Lasanta T, Marti C, Gonzalez C, Errea MP, Valero B. 1997. Assessing the effects of land-use changes on sediment
yield and channel dynamics in the central Spanish Pyrenees. In Human Impact on Erosion and Sedimentation, Walling DE, Probst JL (eds).
International Association of Hydrological Sciences: Wallingford; 151–158.
Gomez B, Eden DN, Peacock DH, Pinkney EJ. 1998. Floodplain construction by recent, rapid vertical accretion: Waipaoa river, New Zealand.
Earth Surface Processes and Landforms 23: 405–413.
Gomez B, Eden DN, Hicks DM, Trustrum NA, Peacock DH,Wilmhurst J. 1999. Contribution of floodplain sequestration to the sediment budget
of theWaipaoa River, New Zealand. In Floodplains: Interdisciplinary Approaches, Marriott S, Alexander J, Hey R (eds). Geological Sciences
of London Special Publication no. 163; 69–88.
Gomez B, Banbury K, Marden M, Trustrum NA, Peacock DH, Hosking PJ. 2003. Gully erosion and sediment production: Te Weraroa Stream,
New Zealand. Water Resources Research 39: 1187. DOI:10.1029/2002WR001342.
Graf WL. 1979. The development of montane arroyos and gullies. Earth Surface Processes and Landforms 4: 1–14.
Griffiths GA. 1982. Spatial and temporal variability in suspended sediment yields of North Island basins, New Zealand. Water Resources
Bulletin 18: 575–584.
Hamilton D, Kelman, EHH. 1952. Soil conservation survey of the Waipaoa River catchment, Poverty Bay, New Zealand. Unpublished Report,
New Zealand Soil Conservation and Rivers Control Council: Wellington, New Zealand.
Harvey AM. 1996. Holocene hillslope gully systems in the Howgill Fells, Cumbria. In Advances in Hillslope Processes, Vol. 2, Anderson MG,
Brooks SM (eds). John Wiley and Sons: Chichester; 731–752.
Harvey MD, Watson CC, Schumm SA. 1985. Gully erosion, U.S. Department of the Interior, Bureau of Land Management, Technical Note no.
366; 181pp.
Henderson J, Ongley M. 1920. The geology of the Gisborne and Whatatutu subdivisions, Raukumara Division. Bulletin no. 21 (new series),
Geological Survey Branch, Department of Mines: Wellington, New Zealand.
Hicks DL. 1991. Erosion under pasture, pine plantations, scrub and indigenous forest: a comparison from Cyclone Bola. New Zealand Forestry
36: 21–22.
Hicks DM, Gomez B, Trustrum NA. 2000. Erosion thresholds and suspended sediment yields, Waipaoa River Basin, New Zealand. Water
Resources Research 36: 1129–1142.
Hill H. 1895. Denudation as a factor of geological time. Transactions and Proceedings of the New Zealand Institute 28: 666–680.
770 M. MARDEN ET AL.
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)
Ireland HA, Sharpe CFS, Eargle DH. 1939. Principles of gully erosion in the piedmont of South Carolina, Technical Bulletin no. 633, U.S.
Department of Agriculture: Washington, DC.
Jones IE, Howie WR. 1970. The measurement and control of erosion and sedimentation. In Proceedings of the New Zealand Water Conference,
Auckland Branch of New Zealand Institute of Engineers, Auckland, New Zealand; 46.1–46.23.
Kelliher FM, Marden M,Watson AJ, Arulchelvam IM. 1995. Estimating the risk of landsliding using historical extreme river flood data. Journal
of Hydrology (NZ) 33: 123–129.
Le Bourdiec P. 1972. Accelerated erosion and soil degradation. In Biogeography and Ecology in Madagascar, Battistini R, Richard-Vindard G
(eds). Junk: The Hague; 227–259.
Liebault F, Piegay H. 2001. Assessment of channel changes due to long-term bedload supply decrease, Roubion River, France. Geomorphology
36: 167–186.
Lyell C. 1849. A Second Visit to the United States of North America, Vol. 2. John Murray: London; 23.
Mazengarb C, Speden IG. (compilers) 2000. Geology of the Raukumara area. Institute of Geological and Nuclear Sciences 1:250 000 geological
map 6. Institute of Geological and Nuclear Sciences: Lower Hutt, NZ; 1 sheetþ 60 pp.
Mazengarb C, Francis DA, Moore PR. 1991. Sheet Y1 Tauwhareparae 1:50 000 scale. Department of Scientific and Industrial Research,
Wellington, New Zealand; 1 sheetþ 52 pp.
NWASCO. 1970. Wise land use and community development. Report of technical committee of inquiry into the problems of the Poverty Bay-
East Coast District of New Zealand. Published for the National Water and Soil Conservation Organisation by the Water and Soil Division,
Ministry of Works: Wellington, New Zealand; 119 pp.þmaps.
O’Loughlin CL. 1974a. The effect of timber removal on the stability of forest soils. Journal of Hydrology (NZ) 13: 121–134.
O’Loughlin CL. 1974b. A study of tree root strength deterioration following clearfelling. Canadian Journal of Forestry Research 4: 107–113.
Parliamentary Commissioner for the Environment 1993. Water and soil resource management on the East Coast. Office of the Parliamentary
Commissioner for the Environment: Wellington, New Zealand.
Patton PC, Schumm SA. 1975. Gully erosion, northwestern Colorado: a threshold phenomenon, Geology 3: 88–90.
Peacock DH, Turner WJ. 2003. Waipaoa River Flood Control Scheme—Proposed Review of Scheme 3B. Gisborne District Council report
GDC2003/410; 150–155.
Pearce AJ, Black RD, Nelson CS. 1981. Lithologic and weathering influences on slope form and process, eastern Raukumara Range,
New Zealand. IAHS Publication no. 132; 95–122.
Pearce AJ, O’Loughlin CL, Jackson RJ, Zhang XB. 1987. Reforestation: on-site effects on hydrology and erosion, eastern Raukumara Range,
New Zealand. In Forest Hydrology and Watershed Management: Proceedings, Vancouver Symposium, August 1987. Publication no. 167,
International Association of Hydrological Sciences: Wallingford; 489–497.
Phillips CJ. 1988. Rheological investigations of debris flow materials. PhD thesis, Lincoln College, New Zealand.
Phillips CJ, Marden M, Pearce AJ. 1990. Effectiveness of reforestation in prevention and control of landsliding during large cyclonic storms. In
Proceedings 19th International Union of Forestry Research Organisations, Montreal; 358–361.
Phillips CJ, Marden M, Miller D. 2000. Review of plant performance for erosion control in the East Coast Region. Unpublished Landcare
Research Contract Report LC9900/111 prepared for Ministry of Agriculture and Forestry, Wellington, New Zealand.
Piegay H, Salvador PG. 1997. Contemporary floodplain forest evolution along the middle Ubaye River, Southern Alps, France. Global Ecology
and Biogeography Letters 6: 397–406.
Piegay H, Walling DE, Landon N, He Q, Liebault F, Petiot R. 2004. Contemporary changes in sediment yield in an alpine mountain basin due to
afforestation (the upper Drome in France). Catena 55: 183–212.
Prosser IP, Abernethy B. 1996. Predicting the topographic limits to a gully network using a digital terrain model and process thresholds,Water
Resources Research 32: 2289–2298.
Prosser IP, Slade CJ. 1994. Gully formation and the role of valley-floor vegetation, southeastern Australia. Geology 22: 1127–1130.
Prosser IP, Soufi M. 1998. Controls on gully formation following forest clearing in a humid temperate environment.Water Resources Research
34: 3661–3671.
Prosser IP, Chappell J, Gillespie R. 1994. Holocene valley aggradation and gully erosion in headwater catchments, south-eastern highlands of
Australia. Earth Surface Processes and Landforms 19: 465–480.
Reid LM, Page MJ. 2002. Magnitude and frequency of landsliding in a large New Zealand catchment. Geomorphology 49: 71–88.
Splus 2002. Splus 6.1 for Windows. Insightful Corporation: Seattle.
Surian N, Rinaldi M. 2003. Morphological response to river engineering and management in alluvial channels in Italy. Geomorphology 50:
307–326.
Trafford JA, 1998. Bedload transport at Kanakanaia, Waipaoa River, East Coast, New Zealand. MSc thesis, University of Auckland.
Wells NA, Andriamihaja B. 1993. The initiation and growth of gullies in Madagascar: are humans to blame? Geomorphology 8: 1–46.
PRE- AND POST-REFORESTATION GULLY DEVELOPMENT 771
Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 757–771 (2005)