Water quality in New Zealand’s planted forests: a review

Physical attributes

Afforestation of a 95 ha catchment in Waikato hill country significantly reduced stream temperatures within six years of planting to temperatures below that considered stressful to sensitive aquatic invertebrates and were predicted to achieve a forested thermal regime within 10 years of planting (Quinn et al. 2009; Quinn et al. 1994). The Blaschke et al. (2008) review of a range of studies across New Zealand found that afforestation was particularly effective in reducing average annual sediment yields in small catchments. Reductions ranged from 50 to greater than 80% although the variability was strongly influenced by underlying geology and rainfall.

Nutrients

The nutrient content of stream water can change substantially after afforestation. For example, total phosphorus (TP) and dissolved reactive phosphorus (DRP) concentrations declined within 4-5 years of planting a small (34 ha) pasture catchment in the Central North Island in P. radiata in comparison with an adjacent pasture stream (Cooper et al. 1987). Nitrate-nitrogen (NO₃-N) response in the afforested catchment varied over time, but long-term records showed a decline in NO₃-N within 5 years of planting (Davis 2014).

Biological attributes

Periphyton biomass declined significantly within six years of afforestation of a 95-ha hill country catchment in the Waikato region of the North Island (Quinn et al. 2009). Macroinvertebrate community composition changed within six years of planting. Increases were observed in the environmentally sensitive taxa (Ephemeroptera (mayflies), Plecoptera (stoneflies) and Trichoptera (caddisflies) (EPT)), reflecting a shift in community structure toward one associated with indigenous forest streams (Quinn et al. 2009). In the same region, Donnison et al. (2004) found that microbial contamination by Escherichia coli declined by a factor of 10 (to a median concentration of 41 E. coli 100 mL⁻¹) over the two years following the afforestation of three small pasture sub-catchments. Some of this effect was attributed to concurrent stock exclusion and pest control. Microbial contamination was also low (median concentration 83 E. coli 100 mL⁻¹) in four small sub-catchments seven years after conversion to Pinus radiata.

Physical attributes

Streams in mature plantation forest streams are highly shaded (mean or median light levels at the water level ranged from 1.3 to 12.1% of an open reference site). Shade levels were strongly influenced by riparian vegetation condition and channel width (Baillie et al. 2005; Boothroyd et al. 2004; Davies-Colley and Quinn 1998; Thompson et al. 2009).

Low light levels in mature planted forest streams limit primary productivity from aquatic plants (i.e. macrophytes and periphyton) (Boothroyd et al. 2004; Thompson et al. 2009). Consequently, temperature is the main factor influencing dissolved oxygen (DO) concentrations (higher concentrations of DO occur at lower temperatures). Dissolved oxygen is an important indicator of water quality and essential to the life supporting capacity of aquatic environments. Hence a minimum concentration of 80% of saturated concentration is required by the Resource Management Act (1991). In most studies, concentrations of DO in mature forests were above this concentration (Collier and Bowman 2003; Collier and Halliday 2000; Matheson et al. 2011; Pruden et al. 1990). An exception of 72% was recorded in a forested Northland stream by Baillie et al. (2005), which the authors attributed to the warmer thermal regimes in this region.

Estimated sediment yields in mature planted forests across New Zealand ranged from 1.7 to 46 t km⁻² yr⁻¹ (Basher et al. 2011; Dons 1987; Fahey and Marden 2006; Hicks 1990). The majority of these sites are in steep hill country. Geology influenced these results with lower yields generated from volcanic soils, Moutere Gravel, and schist. Highest sediment yields were generated in terrains with greywacke parent material and Separation Point Granite. Turbidity and suspended sediment (SS) concentrations in mature planted forests were low across all geological groups (turbidity ≤ 2 NTU’s; SS concentrations < 3 g m⁻³) except where gullying occurred in soft-rock geologies (turbidity 0.8 to 100 NTUs; SS concentrations < 1.2 to 100 g m⁻³). Elevated turbidity and suspended solid concentrations and low water clarity were also recorded in a first rotation planted forest in the Waikato hill country, and were attributed to the re-mobilisation of channel bank material accumulated under a previous pasture land-use regime (Quinn et al. 1997).

Nutrients

Concentrations for a range of N and P attributes measured in mature planted forest streams varied by at least an order of magnitude (Figure 2) but in most studies, concentrations were low. The reasons for the higher levels of nutrients recorded in some studies were not always obvious, but underlying lithologies and associated soils, previous land-use practices and localised inputs of atmospheric N are possible contributing factors (Davis 2014; Macaskill et al. 1997; Parkyn et al. 2006).

Biological attributes

Similar to physico-chemical attributes, biological indicators of water quality in mid-rotation to mature planted forest stands also varied considerably. Concentrations of E. coli in mid rotation and mature planted stands were generally below 150 E. coli 100 mL⁻¹ (Fahey and Stansfield 2006; Parkyn et al. 2006; Young et al. 2005). The most likely source is faecal contamination originating from birds and feral animals such as possums, pigs, and deer.

Effects of management activities

Few published studies in New Zealand have evaluated the effects of forestry management activities (such as silvicultural treatments and pesticide application) on water quality during the growing phase of the forestry cycle and most studies are dated (i.e. Baillie et al. 2015; Collier and Winterbourn 1989; Jackson 1987; Neary and Leonard 1978). Some of these management activities have the potential to affect water quality e.g. Binkley et al. (1999) and Neary (2012).

The land preparation practice of V-blading in West Coast wetlands, where soil was mounded in rows to facilitate drainage and provide elevated sites for planting trees (Collier and Winterbourn 1989; Jackson 1987), elevated summer water temperatures in streams. The authors also recorded higher dissolved aluminium concentrations, and lower pH than streams draining indigenous forest catchments, lower aquatic invertebrate taxa richness in some very acidic streams and an increase in annual sediment yields.

Two studies (Leonard 1977; Neary and Leonard 1978) covering a range of New Zealand sites with varying geology, climate and hydrology found fertiliser application to trees on nutrient-deficient sites resulted in short-term increases in N and P in forest stream waters. The main pulse of N or P was detected either immediately (particularly when application was over the stream channel), just after fertiliser application, or during high flow events in the weeks or months immediately after fertiliser application. In general, nutrient levels returned to baseline levels within a few weeks or months of application. The authors did not consider the levels in stream water to be harmful as only a small portion of the total amount of fertiliser (0.5% of total application) reached the streams.

Some conifers, including P. radiata, are susceptible to the Dothistroma needle blight (Bulman et al. 2004), a disease treated with aerial application of a copper (Cu) fungicide spray. A preliminary study by Fish (1968) found very low levels of Cu (often below detection limits) in stream water and small, but detectable, quantities of Cu in fine-particulate debris. Copper accumulation in aquatic invertebrates was variable, with highest concentrations occurring in the shells of a freshwater snail Potamopyrgus spp.

More recently, research on the aquatic fate of two commonly used herbicides in planted forests (hexazinone and terbuthylazine) showed that under operational conditions, short-term elevated concentrations in stream water occurred either on the day of herbicide application or during a rainfall event shortly after application. Thereafter concentrations were below drinking water standards and lethal concentrations for aquatic organisms and were rapidly diluted downstream, posing a low risk to the aquatic environment (Baillie et al. 2015).

Physical attributes

One of the greatest changes associated with clear-cut harvesting to the stream edge is the increase in light levels (Allan and Castillo 2009). Post-harvest light levels of 11 to 88% (compared with an open reference site) have been recorded in New Zealand streams in the first few years after harvest with marked increases in light occurring at stream widths of around 3.5 to 5 m. Residual shade was provided by stream banks, logging slash, surrounding topography and any remaining vegetation. In some instances, residual logging slash across the stream channel resulted in light levels similar to those found in un-harvested streams (<5% of available light) (Davies-Colley and Quinn 1998; Thompson et al. 2009). Time frames for shade recovery were influenced by stream size and riparian vegetation regrowth (Baillie et al. 2005; Boothroyd et al. 2004; Davies-Colley and Quinn 1998; Thompson et al. 2009).

Temperature is an important factor influencing in-stream processes (Allan and Castillo 2009). Increased solar radiation following harvest to the stream edge often resulted in higher maximum stream temperatures (up to 25 to 30°C) and diurnal temperature ranges of up to 12°C in the warmer northern latitudes of New Zealand and where the hydrology was dominated by surface and sub-surface flow (Baillie et al. 2005; Boothroyd et al. 2004; Graynoth 1979; Quinn and Wright-Stow 2008) (Table 1). Streams with cooler base flow temperatures, overhead logging slash cover and inputs from groundwater seepages, mediated the impact of reduced shade and increased solar radiation on water temperatures after harvest at some sites (Table 1) (Fahey and Stansfield 2006; Pruden et al. 1990). It is likely that, in some streams, temperatures after harvest are exceeding levels and duration periods considered stressful for some sensitive aquatic invertebrates and fish species (≥23°C for aquatic invertebrates and ≥ 28°C for indigenous fish) (Olsen et al. 2012; Quinn et al. 1994; Richardson et al. 1994).

Stream temperature recovery is closely linked to stream size, hydrology and riparian vegetation characteristics. Smaller streams are more thermally sensitive but riparian influence on shade is greater so thermal recovery is usually faster than for larger sized streams (Cristea and Janisch 2007; Moore et al. 2005). This was evident in Coromandel (Waikato region) streams where the time taken for thermal regimes to return to pre-harvest levels was six to eight years in small (2-4 m wide) streams, but predicted to take 12 to 16 years for medium sized (6-12 m) streams (Quinn and Wright-Stow 2008). Similar time frames for thermal recovery have been found in streams in the Pacific Northwest region of the USA (Moore et al. 2005).

The DO concentrations in stream water often decline in association with temperature increases after harvest, although microbial respiration (increased biological oxygen demand from processing residual in-stream organic matter from harvesting) and primary production can also affect DO (Allan and Castillo 2009). While minimal and short-term impacts on DO were observed in some streams, in others DO concentrations declined below the Resource Management Act threshold of 80% of saturated concentration (Resource Management Act 1991), with recovery to pre-harvest levels extending beyond one year (Baillie et al. 2005; Collier and Bowman 2003). Collier & Bowman (2003) did note that high aeration rates contributed to the recovery of DO within 500 m downstream of one site.

Harvesting in steep hill country often increases sediment yields with most sediment reaching waterways via run-off from roading activities, landslides, and channel-bed scouring with minor contributions from slope erosion (Fahey and Marden 2006; Fransen et al. 2001; Marden et al. 2006). For example, there was an eight-fold increase in sediment yields from harvesting in a Hawke’s Bay catchment (Figure 3) and a five-fold increase in sediment yields in Separation Point Granites in Nelson (not graphed), compared with pre-harvest sediment yields (Basher et al. 2011; Fahey and Marden 2006). Sediment yields elevated by harvesting activities typically returned to pre-harvest levels, or declined markedly, within two to six years of harvest (Basher et al. 2011; Fahey and Marden 2006; Phillips et al. 2005).

Graynoth (1979) and Thompson et al. (2009) both measured increases in stream bed sedimentation after harvesting. In the case of Baillie (2006), headwater streams were still impacted by sediment six years after harvest where woody debris and re-generating vegetation retained sediment within the stream channel. Recovery time will depend on shading out and die back of vegetation in the channel caused by the next rotation of trees, decomposition of woody debris and the capacity of the streams to flush sediment from the system. In spite of large increases in sediment yield during the harvest period (Figure 3), Fahey and Stansfield (2006) found no significant changes in water clarity, turbidity and suspended sediment after harvest. One possible explanation is that most suspended sediment was exported during elevated flows (Fahey et al. 2003; Phillips et al. 2005), and was occasionally captured during routine water quality monitoring.

Nutrients

Increases in nutrients, particularly N and P, typically follow clear-cut harvest operations up to the stream edge as a result of disruption to nutrient cycles and increased leaching and sediment transport into water ways (Gundersen et al. 2006; Hartman 2004; Pike et al. 2010). Many studies have been confined to specific regions (Figure 1). For example, Graynoth (1979) recorded average NO₃-N concentrations of 0.12 and 0.36 g m⁻³ in two Nelson streams after clear-cut harvesting to the stream edge compared with the un-harvested control site concentration of 0.05 g m⁻³. The TP concentrations showed little response remaining below 0.05 g m⁻³. Thompson et al. (2009) also recorded an increase in NO₃-N (average of 1.43 g m⁻³) two years after harvesting at four Otago/Southland streams compared with a pre-harvest average of 0.30 g m⁻³. Collier and Bowman (2003) found the effects of harvesting on dissolved organic carbon, NO₃-N and DRP were short-term (recovery within six months of harvesting) in spring-fed central North Island streams. Post-harvest nutrient increases do not occur in all cases. A Hawke’s Bay hill-country catchment showed no significant changes in TP, soluble reactive P, total nitrogen (TN), NO₃-N, or NH₄-N (Ammonium-nitrogen) after harvesting (Fahey and Stansfield 2006) possibly attributable to the BMPs employed during the harvest operation (Eyles and Fahey 2006) and there were no significant increases in NO₃-N after harvest of a small central North Island catchment with a high background concentrations of N (Parfitt et al. 2002).

Biological attributes

The physio-chemical changes in water quality following forest harvest to the stream edge produce a range of biological responses. Increased periphyton production in response to elevated in light, temperature and nutrient concentrations, often peaked in the first two to five years after harvest (Boothroyd et al. 2004; Death and Death 2006; Graynoth 1979; Reid et al. 2010; Thompson et al. 2009). Invertebrate densities and biomass frequently increased after harvest in response to increased primary productivity and in-stream organic matter. However, factors such as warmer and more variable temperature regimes and increases in fine sediment often initiated a shift of aquatic invertebrate community composition to those more closely associated with open pasture sites. Sensitive EPT taxa were replaced by taxa (predominantly Chironomidae, Oligochaetes and Mollusca) more tolerant of disturbances associated with harvesting (Baillie et al. 2005; Collier and Bowman 2003; Death et al. 2003; Graynoth 1979; Quinn et al. 2004; Thompson et al. 2009). A Nelson study also recorded a decline in freshwater crayfish (Paranephrops planifrons) (Graynoth 1979).

Spatial and temporal effects of harvesting

The spatial and temporal distribution of harvesting activities within a planted forest landscape influences both the magnitude of stream ecosystem response and the subsequent recovery rate (Collier and Smith 2005; Reid et al. 2010). Progressive harvesting over a nine to ten year period in two North Island catchments increased the amount of fine sediment, wood and macrophytes in these stream systems along with a decline in the percentage EPT abundance and an increase in the percentage abundance of Elmidae (Collier and Smith 2005). In the Coromandel area, Reid et al. (2010) found the greatest changes to aquatic invertebrate communities following harvest occurred at a threshold catchment size of <500 ha and an upstream harvest area of >40% of the catchment, particularly when clear-cut harvesting to the stream edge. There was a significant decline in QMCI as the proportion of area harvested upstream increased. While invertebrate communities in catchments < 100 ha in size had largely recovered to pre-harvest characteristics within 8 years of harvest, recovery was slower in some larger catchment sites. Restoration of shade and associated thermal regimes are likely factors influencing this recovery rate. Contributing factors such as regional variation in geology and hydrology, previous land-use, and inter-annual variation also influenced aquatic invertebrate response in both these studies (Collier and Smith 2005; Reid et al. 2010).

Conversely, in the Hawke’s Bay region it was the small headwater streams that took longer to recover from harvesting activities. Invertebrate community characteristics were still affected six years after harvest where channels were impacted by sediment deposition and logging slash, whereas harvesting had little effect on channel morphology at the base of the catchment, and invertebrate community characteristics had recovered within four to five years (Baillie 2006; Death and Death 2006).

Climate change forecasts predict an increase in the frequency of extreme weather events with greatest risks associated with harvesting of planted forests in highly erodible and steep topography (Dunningham et al. 2012). Extreme weather events over large spatial scales can override the effects of land use, including harvesting activities (Basher et al. 2011) and are likely to prolong post-harvest recovery of water quality in streams.

Physical attributes

Forest type has little effect on light levels reaching streams with median light levels of <8% measured in both planted and indigenous forest streams compared with median light levels up to 45% for streams flowing through pasture (Boothroyd et al. 2004; Davies-Colley and Quinn 1998; Quinn et al. 1997).

As a result, stream temperatures were usually higher in pasture streams compared with both planted and indigenous forest streams. In the Waikato region, daily mean temperature averaged 15°C in pasture streams, around 2°C higher than in indigenous and planted forest streams. Daily maximum temperatures were 6 to 7°C higher in pasture streams with maximum temperatures often exceeding 20°C. By contrast, water temperatures rarely exceeded 15°C in forested streams. However there was little difference in daily minimum temperatures among the three land uses (Quinn et al. 1997). Similar results were observed in the Gisborne (Parkyn et al. 2006), Hawke’s Bay (Fahey and Stansfield 2006), Nelson (Young et al. 2005) and Canterbury (Harding and Winterbourn 1995) regions.

Nutrients

The range of TN, NO₃-N/NO_x-N (oxidised-nitrogen) and ammoniacal nitrogen (Amm-N)/NH₄-N concentrations in streams in planted forest catchments were similar to indigenous forest streams and generally lower than those in pasture catchments with their more intensive fertiliser and animal stocking regimes (Figure 4a). However, in a central North Island study, the trend was reversed for NO₃-N, with highest concentrations recorded in the indigenous forest site (pasture = 0.013 g m⁻³; planted forest = 0.176 g m⁻³; native forest = 0.805 g m⁻³) (Cooper and Thomsen 1988; Cooper 1986; Cooper and Cooke 1984). The authors attributed this difference to denitrification by vegetation in the pasture and planted forest stream channels and suppression of NO₃-N formation in the planted forest soils influenced by the presence of tree root systems.

Total P concentrations showed a similar land use pattern to TN (Figure 4b). Concentrations of DRP showed little difference between the three land-uses and were below 0.021 g m⁻³ (Figure 4b), although in Matheson et al.’s (2011) study, total DRP concentrations in indigenous forest streams were significantly lower than those in planted forest or pasture streams (0.019, 0.068 and 0.084 g m⁻³ respectively).

Land-use also influenced the timing and forms of nutrient export with N and P mainly transported in dissolved forms during baseflow conditions from a planted forest and indigenous forest catchment with most of the N and P export from the pasture catchment occurring in particulate forms during storm events (Cooper and Thomsen 1988).

Biological attributes

Underlying water quality attributes influenced by land use such as stream lighting, water temperature, sediment, organic matter and nutrients were key factors driving differences in primary productivity, community respiration, and biological community compositions between streams in forested and pasture catchments (Friberg et al. 1997; Harding and Winterbourn 1995; Parkyn et al. 2006; Quinn et al. 1997; Young and Huryn 1999).

Streams flowing through Waikato hill-country pasture had significantly higher gross photosynthetic rates and periphyton biomass than streams in either indigenous or planted forests (Quinn et al. 1997). Similar land-use effects on algal productivity were evident in other regions of New Zealand (Death et al. 2003; Thompson and Townsend 2004). However, in the Otago region, streams in planted forests supported higher algal biomass than those in pasture, tussock or indigenous forests (although this difference was not statistically significant) (Townsend et al. 1997), and in the Canterbury region no significant differences were found in algal biomass between streams in pasture, scrub or forested catchments (Harding and Winterbourn 1995). In the latter study, the author’s suggested that the dense riparian grasses shading the pasture stream channels may have been influencing these results.

Greater concentrations of E. coli are generally found in streams running through pasture land due to animal stocking in the catchment, than streams in forested catchments (Donnison et al. 2004; Parkyn et al. 2006; Young et al. 2005). However, E. coli concentrations in both planted forest and indigenous forests stream did exceed stock drinking and recreational guidelines on certain occasions. The mostly likely sources of bacteria were feral animals and birds.

Streams in pasture catchments often supported higher densities of aquatic invertebrates than those in forested sites (Figure 5a) mainly in response to the higher enrichment and productivity in these streams. Invertebrate communities were characterised by high abundances of more pollution tolerant taxa such as Chironomidae, Mollusca, and Diptera (Death et al. 2003; Friberg et al. 1997; Harding and Winterbourn 1995; Parkyn et al. 2006). The aquatic invertebrate indices in Figure 5(b-e) were generally indicative of improving water quality from pasture to planted forest to indigenous forest. Only streams in indigenous forest catchments consistently had MCI or QMCI scores above the level indicative of ‘clean water’. In some studies, the differences in %EPT taxa, EPT density, MCI and QMCI between pasture and forested streams were significant (Parkyn et al. 2006; Quinn et al. 1997; Shearer and Young 2011).

Land-use comparisons with national water quality standards

Many water quality research publications lack the long-term data sets needed to benchmark the performance of different land-uses against national water quality standards and guidelines. Ballantine et al. (2010) and Larned et al. (2004) used long-term national and regional water quality datasets for these types of analyses and found that TN, NO_x-N, HN₄-N, DRP and E. coli concentrations were significantly higher and water clarity significantly lower in pasture and urban sites compared with planted and indigenous forests (Ballantine et al. 2010; Larned et al. 2004). Pasture and urban sites exceeded national water standards for all these water quality parameters, whereas indigenous forest sites exceeded national water quality guideline for E. coli concentrations and planted forests for conductivity (guideline of 175 μS/cm). Given the influence of geology on conductivity in mature planted forests (Figure 6) and the establishment of planted forests in many areas of New Zealand on steep, highly erodible land, sediment concentrations are a likely factor influencing these results.

National and regional water quality monitoring sites for planted forests are often established in larger catchments where planted forests may be the dominant, but not necessarily the only land use in the catchment which may influence results. Monitoring in larger catchments also limits the ability to detect temporal changes in water quality that occur throughout a forestry cycle at the smaller spatial scale. In contrast, research papers on water quality in planted forests usually focus on smaller catchments entirely in planted forests. These more detailed studies provide information on a wider range of attributes than can be reported on nationally, and capture the temporal and spatial variability inherent in planted forests. Therefore, considering results from both national and research focused studies provides the best depiction of water quality in planted forest streams.

Land-use effects on lakes and wetlands

Lakes and wetlands lack the flushing capacity of streams and rivers and are inherently more susceptible to effects of land management activities. Overall, water quality in lakes across New Zealand is declining (Verburg et al. 2010). Abell et al.’s (2011) survey of 101 lakes across New Zealand showed that while high producing pasture land had the greatest effect on lake nutrient status, concentrations of TP were positively correlated with the percentage of planted forest in the catchment and exceeding the naturally high background rates of P associated with underlying volcanic geology. In addition, most particulate forms of P are usually exported during high flow events, so regular water quality monitoring programmes may underestimate P export from planted forests to downstream environments such as lakes. The authors suggested that the legacy from historical forest clearance and previous land use may also be influencing these results.

The influence of planted forests on the water quality in a range of lotic water bodies (including wetlands) in the Otago region of the South Island was less obvious (Galbraith and Burns 2007). Although water quality from planted forests was more closely associated with pasture and urban areas than indigenous vegetation in ordination space, correlations between nutrient concentrations and the percentage of the catchment in planted forest were either weak or absent. Verburg et al. (2010) also found no significant correlations between the percentage of planted forest land cover and the Trophic Lake Index (TLI) and its components of TN, TP and chlorophyll a. However, in catchments where planted forest formed the dominant land cover, lakes supported higher levels of macrophytes as measured by the submerged plant index (SPI).