Forest Ecology and Management 256 (2008) 949957

Contents lists available at ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier.com/locate/foreco

Vegetation structure, composition and effect of pine plantation harvesting on riparian buffers in New Zealand
E.R. (Lisa) Langer a, G.A. Steward b,*, M.O. Kimberley b
a b

Scion, Forest and Environment, P.O. Box 29 237, Christchurch, New Zealand Scion, Forest and Environment, Private Bag 3020, Rotorua, New Zealand

A R T I C L E I N F O

A B S T R A C T

Article history: Received 22 January 2008 Received in revised form 21 May 2008 Accepted 22 May 2008 Keywords: Riparian vegetation New Zealand Radiata pine Plantation forests Harvesting

The composition and structure of vegetation within riparian buffers prior to, and immediately postharvesting in a managed radiata pine (Pinus radiata D.Don) forest is described and compared with riparian buffers in residual adjoining native forest on the Coromandel Peninsula, New Zealand. One hundred and twenty-one species (71% native) representing life forms from grasses to trees were recorded. The highest species richness, including both native and adventive (non-native) species, was found in riparian buffers in the post-harvest and native reference sites which had 1825 species per site. Riparian buffers in mature pine plantations contained a mix of native species that was generally similar to, and not signicantly reduced in species richness, from the reference native forest. Native species comprised 8292% of the total cover in mature pre-harvest sites (irrespective of riparian width), and 99.8% in native reference sites. Compared with native forest the principal difference was a reduction of total cover in the upper tiers (512 m), and some increase in cover in the lower tiers. Adventive species in post-harvest sites comprised 1667% of the total cover and were most frequently found in riparian areas highly disturbed by recent harvesting of the pines, particularly where riparian buffers were narrow or absent. Invasion by light-demanding adventives is expected to be temporary and most species are likely to be shaded out as the new rotation of pine trees develops. Radiata pine plantations in Whangapoua Forest can provide suitable conditions for the development of riparian buffer zones that will become dominated by native species, similar in richness and structure to neighbouring native forest. 2008 Elsevier B.V. All rights reserved.

1. Introduction Vegetated riparian buffer zones are often advocated as a suitable environmental management tool for reducing impacts of land use activities, including forestry, on aquatic resources by buffering streams from changes in habitat and energy supply (volume and velocity of ow entering the system), and increases in contaminant loads (Gilliam et al., 1992; Collier et al., 1995; Lowrance, 1998; Lowett and Price, 1999). Vegetation is widely accepted as a key factor in stream bank stability. Above ground vegetation is known to increase roughness slowing oodows and trapping ne sediments when the stream channel expands into the riparian zone (Smith, 1976; Platts et al., 1985). Vegetation spreads and divides the incoming overland or channelised ow and reduces the velocity of the water with an associated reduction in water depth resulting from

* Corresponding author. Tel.: +64 7 3435899; fax: +64 7 3480952. E-mail address: greg.steward@scionresearch.com (G.A. Steward). 0378-1127/$ see front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2008.05.052

increased inltration in the undisturbed area (Clinnick, 1985). A vegetative mat protects banks from scouring although excessive riverbank shading by trees or shrubs will inhibit ground cover growth (Collier et al., 1995), resulting in greater sediment inputs (Smith, 1992) and increased bank erosion (Murgatroyd and Ternan, 1983). Below ground, roots increase bank stability by protecting soils against entrainment from oodows, and root mass and density provide soil strength and thereby protect against gravity collapse of undercut banks (Smith, 1976; Kleinfelder et al., 1992). However, riparian vegetation can destabilise banks if inappropriate vegetation types or planting densities are established (Collier et al., 1995). The use of poplars (Populus sp) and willows (Salix sp) for protecting eroding stream banks has been widely practised in New Zealand (e.g., Van Kraayenoord and Hathaway, 1986). Some planted native species have not been found to match the vigour and protection that poplars and willows provide (Collier et al., 1995; Czernin and Phillips, 2005). However, evidence from surveys of rapidly eroding stream banks in pine forests near Hamilton, New Zealand (Smith et al., 1993) suggest that native tree ferns (Cyathea

950

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

sp and Dicksonia sp) with their extensive brous root systems can provide excellent stream bank protection. Collier et al. (1995) note that kanuka (Kunzea ericoides (A.Rich.) Joy Thomps.) and manuka (Leptospermum scoparium (Labill.) J.R. & G.Forst.) appear to have the inherent characteristics required for stream bank protection, but further study is required. Czernin and Phillips (2005) compared the performance of the native Cordyline australis (G.Forst.) Endl. with willows for riverbank protection. Their results for root breakage and pullout indicated that, on its own, Cordyline australis was inferior to willows, however, riparian vegetation combining Cordyline australis and Phormium sp. performed similarly to willow, especially in low-order streams. Marden et al. (2004) compared the effectiveness of 12 native woody species for stream bank and slope stabilisation. All were recognised as early successional species and showed good below ground growth attributes suited to colonising steep and unstable riparian slopes. In a comprehensive study of stream environments in Nelson, New Zealand, Graynoth (1979) found that a 30-m wide riparian buffer of native trees, ferns, and shrubs left alongside each side of the stream channel was effective in reducing the impact of plantation forest logging on the aquatic environment and fauna, in contrast to catchments with no riparian buffer. Wylie (1975) concluded that 30-m wide riparian buffers, with variation in width suited to local conditions, were sufcient to protect stream environments in New Zealand. Erman et al. (1977) (in Clinnick, 1985) found that riparian buffers had to be greater than 30 m to achieve similar channel stability and species diversity to that achieved in unlogged streams in a California (U.S.) study. Boothroyd and Langer (1999) compared international guidelines for riparian buffers in relation to forest harvesting and noted the lack of a xed riparian width in the New Zealand guidelines. However, they found that narrower riparian zones (<10 m) may be adequate to protect streams although no evidence is available in the New Zealand context. If narrow riparian buffers could be shown to be effective, they would also reduce the loss of production. In a study performed in Whangapoua Forest, the productive area of forest affected by roading and riparian areas was calculated to be 14% with 20 m and 28% with 40 m wide riparian strips on either side of stream centre lines, assuming that 75% of total stream lengths were buffered (Visser and McConchie, 1993). If no harvesting or log hauling is allowed in the riparian zone, then the reduction in the harvestable area is compounded by the additional roading and landings which are required. In turn, these will often be in more difcult locations, have higher adverse environmental impacts, and cost more to construct and maintain. The riparian vegetation analysis described in this paper was part of a larger multi-disciplinary study of the inuence of riparian vegetation along afforested streams which included stream geomorphology and periphyton (Boothroyd et al., 2004), macroinvertebrate (Quinn et al., 2004) and sh studies (Rowe et al., 2002). The study took place in plantation forest in Whangapoua Forest, Coromandel Peninsula, New Zealand between 1998 and 2001. The riparian buffers in harvested catchments were found to enhance native sh communities, with greater numbers of banded kokopu (Galaxias fasciatus) in streams with riparian buffers than in those where riparian zones were absent (Rowe et al., 2002). Quinn et al. (2004) showed that clearcut reaches had lower aquatic invertebrate diversity, taxon richness, and relative abundance compared with sites with continuous buffers which did not differ from intact native or mature plantation forest. Logged sites with patch buffers had intermediate biometric values. In this paper, the structure and composition of riparian plant communities in varying combinations of presence and absence of pine canopy are examined and compared with intact native forest.

2. Methods 2.1. Study sites This study was undertaken in four catchments within Whangapoua Forest on the Coromandel Peninsula (Fig. 1). The physical riparian environment is detailed in Boothroyd et al. (2004). In brief, the catchments are characterised by steep topography. Rivers and streams, with highly variable ow rates, drain northward from the Coromandel Ranges (maximum elevation 573 m) into the Whangapoua Harbour. Soils are predominantly yellow-brown earths with clay overlaying weathered greywacke and andesite rocks. Mean annual rainfall is 1700 mm, but rainfall up to 2300 mm per annum has occurred in recent years (C. Nelson pers. comm.). Extreme storm events occur in most years (such as 1520 mm/h in July 2000). Coromandel Peninsula was formerly covered in native forest, dominated by large kauri (Agathis australis (D.Don) Lindl.) that was logged from the early 1800s to the 1930s. Cleared areas were grazed and burned to promote grass growth until the 1940s (Quinn et al., 2004). Large areas of Whangapoua Forest (7560 ha) have been planted with exotic pines (predominantly radiata pine, Pinus radiata) since the early 1960s. Cutover native forest remnants were retained in the headwaters of some of the catchments, and as patches of riparian forest alongside some of the rivers and streams. Areas of native riparian vegetation vary in width from over 20 m (of relatively unmodied native forest) to margins less than 35 m alongside mature pine plantings. All riparian buffers generally contain a diverse array of predominantly native species. Principal native species include hardwood species such as mahoe (Melicytus ramiorus J.R.Forst. & G.Forst.), ve-nger (Pseudopanax arboreus (Murray) Philipson), big leaved Coprosma species such as kanono and karamu (C. grandifolia (Hook.f.), C. robusta Raoul), kanuka, manuka, rangiora (Brachyglottis repanda J.R.Forst. & G.Forst.), Pittosporum colensoi (Hook.f.), nikau palm (Rhopalostylis sapida H.Wendl. & Drude), and tree ferns such as mamaku (Cyathea medullaris (G.Forst.) Sw.), C. dealbata (G.Forst) Sw.), and wheki (Dicksonia squarrosa (G.Forst.) Sw.). Infrequent semi-mature kauri, and other native conifers, also remain and are regenerating. The presence of radiata pine within the riparian buffers appears to be dictated by terrain. Where

Fig. 1. Location of Whangapoua study site in the North Island of New Zealand.

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

951

streams are deeply incised, the upper riparian boundary tends to follow the contours with pine rarely planted down steep dropoffs into streams and rivers. The inverse is true for atter terrain. In Whangapoua Forest, harvesting of pine plantations involves clear cutting, using skyline cable haulers on slopes >128, with heelboom loaders or ground-based operations in at terrain (<128). The operating guidelines ensure earthworks associated with landing and road construction are kept clear of major watercourses. Disturbance to riparian vegetation along streams draining catchments >50 ha is minimised where possible. After harvesting, clearfelled areas are aerially sprayed with herbicide in autumn for weed control, and then oversown with a mix of grasses and legumes (usually within 6 months of harvest). 2.2. Riparian vegetation assessment Vegetation composition, density, and structure were measured using non-bounded reconnaissance vegetation plots (Allen, 1992). In October 1998, in a pilot study to determine appropriate sample sizes, six plots were assessed in one mature pine stand next to the Opitonui River. These plots were re-assessed 5 months later following harvesting in January 1999. A site was dened as a reach of river or stream normally of 100 m length. On the basis of the 1998 pilot study, two plots per site were considered appropriate for the remainder of the study. Two plots per site resulted in a minimum of 10% of the length of the reach being sampled where maximum reach length was 100 m. Within each site, the centres of each plot were placed alternately on either side of the stream to allow for aspect/exposure differences at the 40- and 80-m marks from the downstream point at the start of each 100 m sampling reach. Vegetation greater than 35 m in height within at least 10 m of the stream edge was assumed to inuence the stream environment as channel widths averaged 36 m (Boothroyd et al., 2004). In January to February 1999, 54 plots were assessed; eight plots in four sites in reference native forest, and the remainder in either mature pine forest scheduled for harvesting or recently harvested pine forest. The sites were alongside rivers and streams that varied in catchment area from 17 to 1308 ha, and were 20240 m in elevation. By 2001, most of the mature radiata pine stands in the study area had been harvested, and 18 of the 22 plots in these stands were reassessed. The plots were classied into six class types on the basis of the status of the forest type (native, recently harvested pine, pre-harvest pine), and the width of the riparian vegetation in the plot. The class types used were: recently harvested pine with no riparian vegetation (H0); recently harvested pine with riparian vegetation less than 10 m wide (H < 10); recently harvested pine with riparian vegetation greater than 10 m wide (H > 10), Mature pine with riparian vegetation less than 10 m wide (P < 10); Mature pine with riparian vegetation greater than 10 m wide (P > 10); and Native Forest (N). The number of sites

and plots represented in each site type and other summary information is given in Table 1. 2.3. Vegetation assessment of reconnaissance plots Vegetation was assessed by modifying the standard reconnaissance methods of Allen (1992). All species in the unbounded plot were classied into six height tiers (>25, 1225, 512, 25, 0.32, <0.3 m). The percentage foliage cover of the overall tier and of each species within each tier was estimated in cover classes (<1, 15, 6 25, 2650, 5175, 76100%) modied from the BraunBlanquet cover-abundance scale (Allen, 1992). Aspect, slope, physiography, parent material, surface characteristics and drainage were assessed in each plot. The width of the riparian buffer was measured for each plot from the stream bank to the edge of the pine canopy (dripline) or cut stumps in logged forest. The percentage cover of slash from thinning, harvesting or windthrown trees was estimated in pine sites only. Litter (above ground leaf accumulation and small branches) and soil depths were measured at ve points within all plots (native and pine), with a graduated soil probe, at the plot centre and 5 m above, below and on either side of the plot centre. Following a procedure devised by Brockerhoff et al. (2003) to examine plant communities with regard to life history of the component species, each recorded plant species was categorised according to three different attributes: plant life form, successional stage, and shade tolerance. Plant life form categories were dened as: forbs (broad-leaved herbs), grasses, sedges and rushes, ferns and fern allies, shrubs (woody perennials of lower stature and lacking a well-dened stem); trees (well-dened and usually single stem), and lianes (Brockerhoff et al., 2003). Species were categorised as either pioneer (colonisers of disturbed open sites, usually shade-intolerant), secondary species (seral regeneration), or primary (late seral or climax species that are able to germinate and grow in shade under a closed canopy). Information on the species was sourced from Brockerhoff et al. (2003), Allan (1961), Webb et al. (1988), and Moles and Drake (1999). 2.4. Lighting Lighting within the riparian vegetation was measured by National Institute of Water and Atmospheric Research (NIWA) staff (Boothroyd et al., 2004) using a LAI-2000 canopy analyser (LiCor Inc., Lincoln, NE, USA), with methods described by Davies-Colley and Payne (1998) and Davies-Colley and Quinn (1998). Light readings were taken close to the ground at the 5 litter/soil depth sample points within each reconnaissance plot. Diffuse noninterceptance lighting (DIFN; a measure of the light received on a horizontal plane as a proportion of that from an unobstructed, uniform sky) was calculated as an index of lighting. Two canopy analyser sensors were operated in tandem following the methodology of Davies-Colley and Quinn (1998).

Table 1 Number of sites and plots, mean riparian vegetation width, and time since harvest by site type Forest type Post-harvest Class type H0 H < 10 H > 10 P < 10 P > 10 N Number of sites 9 11 11 7 6 4 Number of plots 19 15 17 12 10 8 Mean riparian width (m) 0 5.4 19.6 3.6 18.5 Mean time since harvest (years) 0.6 1.4 1.4

Pre-harvest

Native

952

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

2.5. Data analysis Mean characteristics for each site type were compared using Analysis of Variance (ANOVA) and Least Signicant Difference (LSD) tests, using SAS Version 9 PROC MIXED (SAS Institute Inc., 2004). In these analyses, the variance between sites within site types was used to test for site type differences. The variables analysed included average numbers of species present per plot by various species classications, and percentage cover. Percentage cover within a height tier was calculated by summing the cover of all species present, using midpoints for each cover class (i.e., 0.5, 3, 15, 37.5, 62.5, 87.5%). When analysing total cover over all tiers of a plant type (e.g., native), cover in each tier was summed without weighting and then adjusted to sum to the total cover for the plot. An arcsine transformation was used in the analysis of percentage data (e.g., percentage cover). As this paper focuses on riparian vegetation in three major site types (native forest, mature pine forest, and recently harvested pine forest) and the effect of riparian width on this vegetation, no attempt was made to match pre- and post-harvest plots in the analysis. An ordination analysis of the species importance values was performed using detrended correspondence analysis (DCA), using the program DECORANA (Hill and Gauch, 1980) as implemented in the analysis package PCORD Version 4.01 (McCune and Mefford, 1999). Cover data in each plot were combined across height tiers to give a single importance value for each species (Leathwick, 1987) with tiers weighted by a modied log10 height scale using tier heights. The percentage variance explained by the DCA axis was estimated using the coefcient of determination between the relative Euclidean distance in the unreduced species space and the Euclidean distance in the ordination space (McCune and Mefford, 1999). 3. Results 3.1. Species richness and diversity The mean number of native species per plot varied from 11.5 in the pre-harvest, narrow riparian buffers (P < 10), and the

harvested sites with no riparian buffers (H0), to 21.8 in the native reference stands (N), and was intermediate in the other site types (Table 2). However, only the extreme values differed signicantly. The number of adventive species per plot differed much more markedly, being highest in the recently harvested pine stands, intermediate in mature pine stands, and extremely low in the native reference stands (Table 2). Total species richness (total number of natives plus adventives) was highest in the harvested pine stands with riparian buffers, and lowest in the mature pine stands with less than 10 m riparian buffer, but again only the extreme values were signicantly different. Native species cover was lowest in harvested stands without riparian buffers where it formed less than half the total cover, intermediate in harvested stands with narrow riparian buffers, with the remaining site types not differing signicantly. 3.2. Riparian vegetation structure Dominant height (mean top height) of vegetation, excluding any pine overstorey, averaged between 4.9 and 6.9 m in the plantation site types, with the exception of the recently harvested stands with no riparian buffer which averaged only 1.1 m in height (Table 3). Dominant vegetation height on the H0 site type was the only one to differ signicantly from the native reference stands, which averaged a little over 7 m in dominant vegetation height (Table 3). Even in the sites with narrow riparian buffers (P < 10 and H < 10), dominant native vegetation height was not signicantly less than the height of native reference stands. Total canopy cover in the recently harvested stands was lower than in the mature pine plantations or the native reference stands, particularly when there was either no riparian buffer or one less than 10 m wide. However, canopy cover in the mature pine stands was similar to the native reference stands, regardless of riparian width (Table 3). In harvested stands with no riparian buffer (H0), there was considerable woody debris or harvesting slash, and greater litter depth, compared with other site types (Table 3). Some windthrow and thinning slash was recorded in the riparian areas of pre-harvested sites, but this only amounted to 10% of the ground cover.

Table 2 Richness of native and adventive vascular plant species (mean number per plot) and native species cover (%) for each site type Forest type Post-harvest Class type H0 H < 10 H > 10 P < 10 P > 10 N Mean number of native species 11.5 b 15.5 ab 17.5 ab 11.5 b 15.8 ab 21.8 a Mean number of adventive species 6.0 a 9.3 a 4.6 ab 1.4 bc 1.8 bc 0.1 c Total number of species 18.4 ab 24.7 a 22.2 a 12.8 b 17.5 ab 21.9 ab Native species cover 33.3 c 63.4 b 83.9 ab 81.6 ab 92.2 a 99.8 a

Pre-harvest

Native

Values within a column followed by the same letter do not differ signicantly (LSD test, p = 0.05).

Table 3 Mean top height, canopy and slash cover and litter depth of all site types Forest type Post-harvest Class type H0 H < 10 H > 10 P < 10 P > 10 N MTH excluding pines (m) 1.1 b 4.9 a 6.3 a 6.2 a 6.9 a 7.1 a Canopy (% cover) 19.3 c 19.8 c 48.8 b 72.4 a 58.7 ab 76.9 a Harvesting slash (% cover) 52.5 a 22.6 b 8.9 bc 10.0 bc 3.3 bc 0.0 c Litter depth (cm) 18.6 a 6.6 b 5.9 b 6.4 b 5.5 b 4.9 b

Pre-harvest

Native

Values within a column followed by the same letter do not differ signicantly (LSD test, p = 0.05).

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

953

Only the mature pine stands with narrow riparian buffers had signicant vegetation above 12 m height and this consisted entirely of the Pinus radiata overstorey (Fig. 2). Cover in the native reference stands was most concentrated in the 512 m height tier.

Cover in this tier did not differ signicantly from pine stands, either pre- or post-harvest with wide riparian buffers (H > 10 and P > 10), but was signicantly lower in stands with narrow riparian buffers (H < 10 and P < 10), and completely absent in harvested

Fig. 2. Mean vegetation cover (%) by tier in each of the six major site types. Each mean cover value is divided into adventive species (dark bars) and indigenous species (light bars).

954

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

Table 4 Percentage of total cover of major plant life forms for each site type Forest type Post-harvest Class type H0 H < 10 H > 10 P < 10 P > 10 N Grasses 22.4 a 19.2 a 13.9 a 3.6 a 4.3 a 1.9 a Forbs 20.1 a 22.1 a 9.1 b 1.9 b 14.0 ab 1.2 b Sedges and rushes 2.0 ab 2.7 ab 1.4 b 2.6 ab 3.9 a 2.6 ab Lianes 35.8 a 7.1 b 5.1 b 1.6 b 1.7 b 1.8 b Ferns 6.9 c 17.1 bc 26.3 ab 39.4 a 27.2 ab 26.8 ab Shrubs 11.0 b 21.9 ab 24.9 a 20.6 ab 20.0 ab 29.2 a Trees 8.4 c 10.8 c 18.9 bc 30.5 ab 27.9 ab 36.4 a

Pre-harvest

Native

Cover values are calculated using equal weighting across tiers. Values within a column followed by the same letter do not differ signicantly (LSD test, p = 0.05). Table 5 Percentage of total cover of species of each successional stage for each site type Forest type Post-harvest Class type H0 H < 10 H > 10 P < 10 P > 10 N Pioneer species (%) 45.3 a 43.0 ab 22.1 b 22.9 ab 10.1 b 11.4 b Secondary species (%) 49.7 a 42.0 a 49.7 a 34.9 a 53.9 a 50.5 a Primary species (%) 7.5 c 16.5 b 29.0 ab 44.4 a 38.3 a 38.1 a

Pre-harvest

Native

Cover values are calculated using equal weighting across tiers. Values within a column followed by the same letter do not differ signicantly (LSD test, p = 0.05).

stands with no riparian buffers (H0). For harvested stands, cover was generally highest in the lower two tiers (<2 m), but no statistically signicant difference was detected. 3.2.1. Vegetation life form structure and succession In the native reference stands and the mature pine stands regardless of riparian width cover consisted predominately of tree, shrub and fern species (Table 4). In contrast, in the harvested pine stands, especially those without riparian buffers, there was a higher percentage of cover distributed among forbs and grasses. The harvested stands with no riparian also had high levels of cover in lianes, although this was entirely due to the presence of the adventive bindweed (Calystegia sp). Lianes were also found in mature pine and native stands. The high incidence of adventives in H0 and H < 10 noted previously (Table 2, Fig. 2) was primarily due to the presence of Calystegia sp, and also to the dominance of adventive forbs and grasses in these sites, and the relatively lower abundance of ferns, shrubs and trees which at all sites were almost all native. The proportion of primary shade tolerant species was higher in native and pre-harvest site types and lower in the harvested pine stands with <10 m riparian buffer (H0 and H < 10) (Table 5). Conversely, there were more pioneer light-demanding species in harvested areas with <10 m riparian buffer (H0, H < 10). The proportion of cover of light-demanding species was closely related to the measured light level (Table 6). Light level was greatest within harvested forest sites with no vegetated riparian buffer (H0,
Table 6 Cover of light-demanding species and DIFN value for each site type Forest type Class type H0 H < 10 H > 10 P < 10 P > 10 N Cover of light-demanding species (%) 75.0 a 57.4 ab 39.7 bc 38.6 bc 26.7 c 32.9 bc Mean DIFN (%) 59.9 a 39.0 b 6.7 c 1.1 c 1.2 c 1.1 c

median DIFN = 60.25% at surface water level, 82.1% at bank level, and 50.8% at ground level). In contrast, very low lighting levels occurred in mature pine forests with a vegetated riparian buffer (P < 10, P > 10) and native reference sites at stream and bank levels (e.g., P > 10 stream median DIFN = 1.75% and bank median DIFN = 2.4%). Ground-level lighting was lowest in mature plantations with less than 10 m riparian buffer (P < 10) (median DIFN = 1.0%). 3.3. Detrended correspondence analysis The rst axis of the detrended correspondence analysis explained 18.7% of the variation, and varied signicantly between site types, being highest in the harvested sites with <10 m riparian buffer (H0 and H < 10), lowest in the native reference stands, and intermediate in the pre-harvested sites (Table 7, Fig. 3). The primary axis was clearly strongly related to plant succession, canopy cover, and shade tolerance, with pioneer species having highest values, primary species lowest values, and secondary species intermediate values (Fig. 4). The primary axis was also higher for light-demanding species (mean 205) than shade tolerant species (49), and higher for adventives (mean 293) than natives (mean 65). Axis 1 scores were signicantly correlated with lighting. The 2nd and 3rd axis did not differ greatly between site types (Table 7), and in combination, explained only an additional 6.5% of variation. This suggests that apart from the strong successional trend associated with light availability described by

Table 7 Mean values of the rst three Decorana axes by site type Forest type Post-harvest Class type H0 H < 10 H > 10 P < 10 P > 10 N Axis 1 201 a 191 a 127 b 94 b 77 bc 29 c Axis 2 113 a 149 a 145 a 131 a 120 a 134 a Axis 3 98 ab 103 ab 122 a 67 b 97 ab 129 a

Post-harvest

Pre-harvest

Pre-harvest

Native

Native

Values within a column followed by the same letter do not differ signicantly (LSD test, p = 0.05).

Values within a column followed by the same letter do not differ signicantly (LSD test, p = 0.05).

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

955

closely resembles that of the reference native forest. Virtually all of the species found in the reference native forest are present in the riparian buffers of the pine plantations. 4. Discussion This study shows that in the Coromandel Peninsula, vegetated riparian buffers along streams and rivers in mature pine plantations maintain a predominantly native vegetation composition not greatly different to neighbouring native forest. The principal differences are a reduction of total cover of native species in the 512 m tier, and an increase of adventive species in the lower tiers. The native species composition in riparian buffers is generally similar and not signicantly reduced in species richness, compared with reference native forest. It is noticeable that, even in mature pine stands with narrow riparian buffers less than 10 m in width, species richness and structure is not greatly different from neighbouring native forest, other than for a generally reduced level of cover. This result is despite the fact that many of these sites were probably completely cleared at the time of plantation establishment, typically about 25 years earlier, and that the existing riparian vegetation is therefore an understorey that has largely developed since forest clearing. It can therefore be concluded that pine plantations in this region can provide good conditions for the development of native understorey vegetation not greatly inferior in richness and structure to neighbouring native forest. This nding is in agreement with other studies that have found a relatively rich understorey of mostly native species in pine plantations in many sites throughout New Zealand (Allen et al., 1995; Ogden et al., 1997; Brockerhoff et al., 2003). However, as noted by Brockerhoff et al. (2003), some long-lived native tree species may be unable to reach full maturity in the 2530 years of a typical pine plantation rotation. Retention of riparian buffers may therefore enable biodiversity within the forest to be retained, in addition to providing other environmental or non-timber benets (i.e., improved soil and water quality, provision of habitats and wildlife corridors, etc.). The native forests on the Coromandel Peninsula were classied into a number of forest classes and types. The principal native forest type within Whangapoua Forest was kaurisoftwood hardwood mixes (Nicholls, 1971). The main long-lived species in this forest type include kauri, taraire (Beilschmiedia taraire (A.Cunn.) Benth. & Hook.f.), rata (Metrosideros robusta) (A.Cunn.), tawa (Beilschmiedia tawa (A.Cunn.) Kirk), hinau (Elaeocarpus dentatus (J.R.Forst. & G.Forst.) Vahl), rewarewa (Knightia excelsa ll.Hall.) R.Br.), rimu (Dacrydium cupressinum Lamb.), miro (Mu (Prumnopitys ferruginea (D.Don) de Laub.), Halls totara (Podocarpus hallii Kirk) and tanekaha (Phyllocladus trichomanoides D.Don). Within the native reference sites six of these species are found with rewarewa being the most common and with the greatest summed cover value. Its presence is expressed in mature individuals and frequent seedlings. Tawa, hinau, rata and tanekaha were also present in native reference sites although at much lower summed cover values and were largely represented either by single mature individuals or few seedlings. Miro was recorded as a single presence only. Kohekohe (Dysoxylum spectabile (G.Forst.) Hook.f.) is not included in this forest type, but was found in large numbers as seedlings and isolated mature trees in the native reference sites. Of these long-lived species only rewarewa and kohekohe were found in moderate amounts, principally as seedlings, in harvested (H > 10) and mature pine (P > 10) riparian buffers greater than 10 m in width. The remaining species were recorded as a presence only, generally in only one site type. The riparian buffers located within mature pine stands show greater abundance of various adventive species, although these are

Fig. 3. Box plots showing the means of primary DCA axis scores within each site type.

the primary axis, there are few other strong species groupings or associations in the riparian buffer and neighbouring native forest vegetation in this study area. 3.4. Primary species in riparian buffers and in reference native forest A number of tree species are dominant in the native reference ll.Hal.) R.Br., nikau forest including mahoe, Knightia excelsa (Mu palm, and Hedycarya arborea (J.R.Forst & G.Forst.), with several shrub species dominating the understorey including rangiora and kanono, along with ferns such as Cyathea dealbata and wheki. These same native species also form the most important element in the riparian vegetation in mature pine plantations (apart from the presence of radiata pine itself, which is evident in sites with narrow riparian buffers), and they are also an important element in riparian buffers of recently harvested stands. These species are also present as small plants in recently harvested stands without riparian buffers indicating that they regenerate adequately on riparian disturbed sites. The principal difference in terms of species composition between the riparian buffers of recently harvested pine stands and native stands is the presence in the lower tier of the pine stands of adventive forbs such as Dianella nigra (Colenso), Lotus pedunculatus (Cav.), Phytolacca octandra (L.), and Senecio jacobaea (L.), adventive grasses such as Holcus lanatus (L.), and Dactylis glomerata (L.), and the adventive liane bindweed (Calystegia sp). These species are also mostly present in riparian buffers of mature pine stands although they have decreased greatly in importance, presumably due to the shading effect of the canopy which more

Fig. 4. Box plots showing the means of primary DCA axis scores of species by successional stage.

956

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

entirely conned to the lower height tiers, compared to reference native forest where such adventive species are almost entirely absent. However, adventive species are a much more important element in riparian buffers of recently harvested pine stands, especially in those that are narrow, and are important in riparian areas where buffers are absent. These adventives which are mainly light-demanding pioneer species become much less important over time as the canopy thickens and rises leading to greatly reduced light levels. As noted by Brockerhoff et al. (2003), the succession of understorey vegetation following harvesting of pine stands is largely driven by changes in light availability. This trend is strongly evident in the riparian buffers in this study, especially when no riparian buffers or only narrow riparian buffers are retained. Even when riparian buffers greater than 10 m width are retained, invasion of light-demanding species occurs following harvesting, presumably because of increased light levels due to damage or die-back of the riparian vegetation, and to the removal of large adjacent pine trees. This invasion of light-demanding adventives is temporary and the species are largely expected to be shaded out as the new rotation of pine trees develops. Trees within narrow riparian buffers are often subject to increased levels of windthrow which may impair some buffer functions. Grizzel and Wolff (1998) found that windthrow affected on average a third of trees in riparian buffers 13 years after clearcut harvest of adjacent timber in 40 riparian buffers bordering small streams in northwest Washington, Rollerson and McGourlick (2001) found that windthrow in riparian buffers tends to increase with increases in rooting depth, wind exposure and tree height. Feathered edges tended to experience less windthrow than straight edges. Narrow riparian buffers (510 m) and 2-sided strips (strips of timber bounded by clearcut areas on either side) were more prone to windthrow, especially when thinned (Rollerson and McGourlick, 2001). However, windthrow was not found to be related to riparian width or thinning in 25 riparian areas in Canada (Ruel et al., 2001). Observations from the current study suggest these patterns may be repeated at sites such as Whangapoua Forest. Incidence of windthrow of riparian vegetation at Whangapoua Forest was most common where newly exposed riparian vegetation was dominated by tall (>57 m in height) and slender stems. Frequent examples occurred in the dense Melicytus ramiorus regrowth, which is a common hardwood species throughout the riparian buffers in Whangapoua. As a consequence of the steepness of the terrain within Whangapoua Forest, the original pine planting boundaries tended to be at the top of steep drop-offs into major watercourses. Isolated peninsulas on the inside of tight stream bends on atter ground have also been created. The resulting sites are often considerably variable in width and frequently were not cleared for planting. Therefore, these riparian buffers often contain remnants of the original native forest cover which existed before pine establishment. Where these remnants were surrounded by mature pine forest they appeared to be healthy and robust. However, as the surrounding pine is felled, these native remnants may degrade over time due to inadvertent or unavoidable damage by harvesting and exposure (including desiccation and greatly increased light levels). It is likely that the species composition within these sites will be reduced as the most sensitive species disappear. However, those species most sensitive to the modication of the riparian may not signicantly impact the overall effectiveness of the riparian buffer because they are typically either minor components (e.g., small ferns) or occupy very selected niches (e.g., epiphytic) within the riparian buffer. To aid the development and health of riparian buffers, some consideration may be required to identify, protect or enhance selected sites throughout the forest to act as local seed sources.

Where the pine forest has been recently felled and the riparian buffer is narrow, or has been signicantly impacted or removed entirely it is usual that the rst wave of regeneration of new riparian species on heavily disturbed sites is almost entirely comprised of adventive weed species. This rst wave is often dense but ultimately of low stature (<3 m). In the short term, the density of this regrowth should be effective in trapping sediment and preventing the movement of debris during extreme storm events but will provide only limited shading effect on rivers and streams. Therefore, it is likely that where seed sources of native shrub species are distant from recently disturbed riparian buffers the development of a tall semi-permanent cover may be delayed. On some recently harvested sites, it was signicant that where native shrub species in the riparian vegetation had been crushed during harvesting operations, but where some stem and root systems remained, it was common for coppice regrowth to be abundant. The regrowth from coppice shoots is often rapid and dense and will often quickly restore a semblance of a native cover. The species most commonly observed with this regeneration strategy were mahoe, rangiora, karamu and the tree ferns. Quinn et al. (2004), in a study of invertebrate communities, concurrent with this study, indicated that riparian buffers in Whangapoua Forest with a mean 18-m width greatly reduced the impact of logging on stream habitat, while Meleason and Quinn (2004) found that riparian buffers in Whangapoua Forest as narrow as 5-m wide substantially moderated air temperature compared to tree-less environments. These and other studies suggest that there are likely to be variations in recommendations for riparian buffer width, depending on the habitat and values being managed. Our current study found that the vegetation composition within mature pine riparian buffers in plots of 10-m width was similar to native reference sites while narrow or completely removed riparian buffer vegetation results in at rst, more light-demanding, usually adventive species. As riparian buffers are modied by harvesting of pines (either within or adjacent to riparian buffers) there are some changes to the residual vegetation in the buffers, albeit initially temporary. Temporal changes to the vegetation within the riparian buffers in Whangapoua Forest are continuing to be monitored and will be reported on in future, which will allow better understanding of suitable riparian widths for the purpose of maintaining the integrity of riparian vegetation. One of the greatest dangers to the long-term survival and health of riparian buffers retained during harvesting is likely to be where management practice prescribes aerial application of herbicides for establishment of the new pine crop. Most if not all native species present in the riparian buffers in Whangapoua Forest are susceptible to the common herbicides used in forest establishment. In Whakarewarewa Forest, native riparian buffers with similar species compositions to those in Whangapoua Forest, were retained as buffers after harvesting as part of the Rotorua District Council Efuent Disposal Scheme but suffered signicant damage from pre-planting aerial application of herbicides (G.A. Steward pers. obs.). Acknowledgements The authors thank Ernslaw One Ltd., especially Chris Nelson, for access to sites, information, and support during eld assessments. We thank colleagues in NIWA, particularly John Quinn for identifying sites, and Kerry Costley for the light data. Chris Ecroyd assisted in allocating life form classication. We thank Brenda Baillie, Murray Davis and one anonymous reviewer for useful comments on earlier drafts. The study was funded as part of the Sustainable Management of Forest Ecosystems Programme

E.R. (Lisa) Langer et al. / Forest Ecology and Management 256 (2008) 949957

957

(contract CO4X0012), by the Foundation for Research, Science and Technology. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.foreco.2008.05.052. References
Allan, H.H., 1961. Flora of New Zealand, vol. I. Government Printer, Wellington, 1085 pp. Allen, R.B., 1992. RECCE an inventory method for describing New Zealand vegetation. FRI Bull. No. 181, Ministry of Forestry, New Zealand. Allen, R.B., Platt, K.H., Coker, R.E.J., 1995. Understorey species composition patterns in a Pinus radiata D. Don plantation on the central North Island Volcanic Plateau, New Zealand. NZ J. For. Sci. 25, 301317. Boothroyd, I.K.G., Langer, E.R., 1999. Forest harvesting and riparian management guidelines: a review. NIWA Technical Report 56, National Institute of Water and Atmospheric Research Ltd., 58 pp. Boothroyd, I.K.G., Quinn, J.M., Langer, E.R., Costley, K.J., Steward, G.A., 2004. Riparian buffers mitigate effects of pine plantation logging on New Zealand streams. 1. Riparian vegetation structure, stream geomorphology and periphyton. Forest Ecol. Manage. 194, 199213. Brockerhoff, E.G., Ecroyd, C.E., Leckie, A.C., Kimberley, M.O., 2003. Diversity and succession of adventive and indigenous vascular understorey plants in Pinus radiata plantation forests in New Zealand. Forest Ecol. Manage. 185, 307326. Clinnick, P.F., 1985. Buffer strip management in forest operations: a review. Aust. For. 48 (1), 3445. Collier, K.J., Cooper, A.B., Davies-Colley, R.J., Rutherford, J.C., Smith, C.M., Williamson, R.B., 1995. Managing Riparian Zones: A Contribution to Protecting New Zealands Rivers and Streams: Vol. 1: Concepts. Department of Conservation, Wellington, NZ. Czernin, A., Phillips, C.J., 2005. Below-ground morphology of Cordyline australis (New Zealand cabbage tree) and its suitability for river bank stabilisation. NZ J. Bot. 43, 851864. Davies-Colley, R.J., Payne, G.W., 1998. Measuring stream shade. J. N. Am. Benthol. Soc. 17 (2), 250260. Davies-Colley, R.J., Quinn, J.M., 1998. Stream lighting in ve regions of North Island, New Zealand: control by channel size and riparian vegetation. NZ J. Mar. Freshwater Res. 32, 591605. Erman, D.C., Newbold, J.D., Ruby, K.B., 1977. Evaluation of streamside buffer strips for protecting aquatic organisms. California Water Resources Institute. Contribution No. 165, Univ. of Calif., Davis, CA, 48 pp. Gilliam, J.W., Schipper, L.A., Beets, P.N., McConchie, M., 1992. Riparian buffers in New Zealand forestry. NZ For. 37, 2125. Graynoth, E., 1979. Effects of logging on stream environments and faunas in Nelson. NZ J. Mar. Freshwater Res. 13 (1), 79109. Grizzel, J.D., Wolff, N., 1998. Occurrence of windthrow in forest buffer strips and its effect on small streams in Northwest Washington. Northwest Sci. 72 (3), 214223. Hill, M.O., Gauch, H.G., 1980. Detrended correspondence analysis, an improved ordination technique. Vegetation 42, 4758. Kleinfelder, D., Swanson, S., Norris, G., Clary, W., 1992. Unconned compressive strength of some streambank soils with herbaceous roots. Soil Sci. Soc. Am. J. 56, 19201925. Leathwick, J.L., 1987. Waipapa ecological area: a study of vegetation pattern in a scientic reserve. Forest Research Institute Bulletin 130, Ministry of Forestry, Forest Research Institute, Rotorua, NZ, 82 pp. Lowett, S., Price, P., 1999. Riparian land management technical guidelines, vol. 1. In: Principles of Sound Management, Land and Water Resources Research and Development Corporation, Canberra, Australia.

Lowrance, R., 1998. Riparian forest ecosystems as lters for nonpoint-source pollution. In: Pace, M.L., Groffman, P.M. (Eds.), Successes, Limitations, and Frontiers in Ecosystem Science. SpringerVerlag, New York, pp. 113141. Marden, M., Rowan, D., Phillips, C.J., 2004. Stabilising characteristics of New Zealand indigenous riparian colonising plants. In: Proceedings of Eco-Engineering: The use of Vegetation to Improve Slope Stability, Thessaloniki, Greece, 1317 October 2004. McCune, B., Mefford, M.J., 1999. Multivariate analysis of ecological data, Version 4. MjM Software, Gleneden Beach, Oregon. Meleason, M.A., Quinn, J.M., 2004. Inuence of riparian buffer width on air temperature at Whangapoua Forest, Coromandel Peninsula, New Zealand. Forest Ecol. Manage. 191, 365371. Moles, A.T., Drake, D.R., 1999. Potential contributions of the seed rain and seed bank to regeneration of native forest beneath plantation pine. NZ J. Bot. 37, 8393. Murgatroyd, A.L., Ternan, J.L., 1983. The impact of afforestation on stream bank erosion and channel form. Earth Surf. Process. Landforms 8, 357369. Nicholls, J.L., 1971. Ecological survey of New Zealands indigenous forests. Type Map Series No 6., Sheet No. 3, Coromandel. Forest Research Institute, New Zealand Forest Service. Ogden, J., Braggins, J., Stretton, K., Anderson, S., 1997. Plant species richness under Pinus radiata stands on the central North Island volcanic plateau, New Zealand. NZ J. Ecol. 21, 1729. Platts, W.S., Gebhardt, K.A., Jackson, W.L., 1985. The effects of large storm events on basin-range riparian stream habitats. Proceedings of a Symposium: Riparian ecosystems and their management: Reconciling conicting uses. US Department of Agriculture and Forest Service Technical Report RM-120, pp. 30 34. Quinn, J.M., Boothroyd, I.K.G., Smith, B.J., 2004. Riparian buffers mitigate effects of pine plantation logging on New Zealand streams. 2. Invertebrate communities. Forest Ecol. Manage. 191, 129146. Rollerson, T., McGourlick, K., 2001. Riparian windthrowNorthern Vancouver Island. In Mitchell, S.J., Rodney, J., et al. (Eds.), Windthrow Assessment and Management in British Columbia, Proceedings of the Windthrow Researchers Workshop, January 31February 1st, Vancouver, University of British Columbia and Forestry Continuing Studies Network, pp. 139155. Rowe, D.K., Smith, J.A., Quinn, J., Boothroyd, I., 2002. Effects of logging with and without riparian strips on sh species abundance, mean size, and the structure of native sh assemblages in Coromandel, New Zealand, streams. NZ J. Mar. Freshwater Res. 36, 6779. Ruel, J.-C., Pin, D., Cooper, K., 2001. Windthrow in riparian buffer strips: effect of wind exposure, thinning and strip width. Forest Ecol. Manage. 143, 105113. SAS Institute Inc., 2004. Whats New in SAS 9.0, 9.1, 9.1.2, and 9.1.3. SAS Institute Inc., Cary, NC. Smith, C.M., 1992. Riparian afforestation effects on water yields and water quality in pasture catchments. J. Environ. Quality 21, 237245. Smith, D.G., 1976. Effect of vegetation on lateral migration of anastomosed channels of a glacier meltwater river. Geol. Soc. Am. Bull. 87, 857860. Smith, M.J., Cooper, A.B., Quinn, J.M., 1993. Land-water interactions research at Whatawhatathe start. NIWA Ecosystems Publication No. 4, National Institute of Water and Atmospheric Research, Hamilton. Van Kraayenoord, C.W.S., Hathaway, R.L., 1986. Plant materials handbook for soil conservation. Vol. 1: principles and practices. And vol. 2: plant materials. Water and Soil Miscellaneous Publication No. 94 Water and Soil Directorate, Ministry of Works and Development, Wellington. Visser, R., McConchie, M., 1993. The impacts of riparian buffer strip characteristics on forest harvesting. LIRO Report 18 (10), 10. Webb, C.J., Sykes, W.R., Garnock-Jones, P.J., 1988. Flora of New Zealand, Volume IV: Naturalised Pteridophytes, Gymnosperms, Dicotyledons. Christchurch, New Zealand. Botany Division, Department of Scientic and Industrial Research, 1565 pp. Wylie, R.E.J., 1975. Forest operations to safeguard water and soil resources. National and Soil Conservation Organisation. Proc. of Water and Soils Division, Course No. 2, Rotorua.