Landscape evolution and hydrogeochemical characteristics of the Pourewa Stream catchment, lower North Island, New Zealand

ABSTRACT

Issues regarding water quality and supply have been raised within the Pourewa catchment, following ongoing landuse change and associated demand for freshwater. Nitrogen loss from farms due to leaching of nitrate into the subsurface has increased in conjunction with landuse intensification. The stream dissects Whanganui Basin fill, including alternating coarse and fine-grained sediments deposited throughout Quaternary climatic cycles, resulting in interbedded aquifers and aquitards in a geologically young landscape. The Rangitikei River dominates the landscape, forming a series of aggradational terraces. Fault lines and basement highs provide a control on both the underlying Plio-Pleistocene strata and modern-day drainage network. A 1:25 000 hydrogeological map and corresponding structural interpretation are presented. Analysis of catchment lithologies demonstrates the importance of grain size, porosity and age in dictating hydraulic conductivity. Organic carbon, illite and glauconite present within the succession indicate attenuation of nitrate within the local groundwater system is taking place.

KEYWORDS:

• Plio-Pleistocene
• stratigraphy
• hydrogeology Whanganui Basin
• aquifer

Introduction

Agriculture constitutes 90% of landuse in the Rangitikei District (Zarour 2008), placing pressure on fresh water resources required for ongoing agricultural development and production. Recent concerns around the availability and long-term sustainability of freshwater resources have initiated research into the districts groundwater systems (Roygard and Carlyon 2004; Begg et al. 2005; McArthur et al. 2007; Wilson 2007; Collins et al. 2017). Limited knowledge and information is currently available on local-scale hydrogeological settings within the Rangitikei District. A sound understanding of local hydrogeology is necessary for improved mapping and quantification of the districts groundwater resources. Furthermore, information on local hydrogeological settings is essential for improving our understanding and ability to map and quantify water and nutrient flow pathways and attenuation from farmlands to receiving waters, informing practices and policies for targeted and effective water quality management. Previous studies have taken a broad approach examining classification, allocation, management and vulnerability of groundwater resources (Roygard and Carlyon 2004; McCarron and Zarour 2005; McArthur et al. 2007; Wilson 2007; Zarour 2008; Callander and Thomas 2013; Morgenstern et al. 2015; Singh et al. 2017). There is a present knowledge gap concerning the hydrogeochemical characteristics of the various rock types within the Rangitikei District. This study addresses that knowledge gap in a single sub-catchment of the Rangitikei.

Our approach centres on geological mapping and developing a conceptual understanding of basin history and landscape evolution. This forms the rationale behind our sampling design and targeted hydrogeochemical analysis of representative catchment lithologies. Although the Pourewa catchment is the focus of this study, we consider a wider perspective is required to both effectively explore and document catchment development through time and understand how regional geology and structure influence the local groundwater system. For these reasons, the geological mapping boundary covers a 1,600-km² portion of the central Rangitikei District centred around the Pourewa Stream (Figure 1). A particularly important factor considered whilst setting the field area was the need to capture and explore the adjacent Rangitikei River, a feature that has played a significant role during landscape evolution.

Figure 1. Location map of the study area displaying the outline of the Pourewa Stream catchment in yellow and relevant place names along its length. Sample locations: S1–Waitapu Shell Conglomerate, S2-Rewa Pumice, S3-Rangitikei River gravel, S4-Pourewa Stream gravel, S5-Potaka Pumice, S6-Mangamako Shellbed, S7-Mangarere Siltstone, S8-Te Rimu Sandstone, S9-Mangamako Sandstone, S10-Hautawa Shellbed, S11-Mangaweka Mudstone. Inset displays location of the study area within Whanganui Basin (WB) and its location relative to the Taupo Volcanic Zone (TVZ), North Island Dextral Fault Belt (NIDFB) and the Australian–Pacific plate interface.

This paper presents data from a range of analyses, providing information on the ability of local lithologies to store and transmit water. We relate organic carbon together with structural and exchangeable ferrous iron (Fe²⁺) to environments of deposition, providing evidence for the subsurface reduction of nitrate within the Pourewa catchment. Evolution of the landscape is tied to eastward migration of the Rangitikei River, tectonics and climatically controlled, Late Quaternary aggradation. The occurrence of freshwater springs are related to geological structure, paleo-channels within the Late Quaternary valley fill and basin architecture, providing evidence for confined aquifers within the study area.

The Pourewa Stream is a c. 35 km long tributary of the Rangitikei River, located in the central Rangitikei District, Lower North Island, New Zealand (Figure 1). The stream drains a 170-km² catchment that begins in steep hill country northeast of Hunterville forming a shallow gradient stream channel before meandering out onto the Rangitikei terrace country at Rata. The stream then continues in a SE direction adjacent to the Rangitikei River before eventually cutting down through the underlying Plio-Pleistocene sediments to its confluence at Onepuhi (Figure 1).

Study area

The islands of New Zealand are but the surface expression of a large submerged subcontinent named ‘Zealandia’. This subcontinent sits astride the convergent margin between the Pacific and Australian lithospheric plates (Figure 1). Tectonic convergence is accommodated by subduction of the Pacific Plate and deformation of the overriding Australian Plate, resulting in active uplift of the North Island (Eberhart-Phillips et al. 2008). Uplift is concentrated along the North Island Dextral Fault Belt (NIDFB) trending roughly parallel to the Hikurangi subduction zone. Uplift along the NIDFB has resulted in the formation of the North Island’s main axial ranges, separating the forearc region from the back-arc and volcanic-arc regions. The Whanganui Basin (WB) is situated in the back-arc region behind the plate interface, southwest of the Taupo Volcanic Zone (TVZ) (Figure 1).

The WB is an approximately 200 × 200 km sedimentary basin located in the south-western North Island of New Zealand (Anderton 1981) containing a c. 4.5 km thick Late Miocene-Pleistocene fill. Basin formation is attributed to lithospheric loading in combination with compressional downwarping driven by coupling of the overriding and subducting plates (Stern et al. 1992, 1993; Tozer 2013). Southward migration of the basin depocentre through the Plio-Pleistocene occurred in combination with uplift along the northern and eastern margins, resulting in present day onland exposures of gently dipping marine to marginal marine strata. Well-preserved fossil Mollusca, foraminifera and nannofossils are used to subdivide and date the basin succession and provide important paleo-environmental information (Beu et al. 2004; Beu 2006). Proximity to the TVZ has resulted in periodic influxes of volcanic products over the last c. 2 Ma (Houghton et al. 1995; Briggs et al. 2005). These volcanic beds establish isochrons used to date and correlate between basin successions (Naish et al. 1996; Pillans et al. 2005).

The most complete sequence of river terraces within the WB occurs in the Rangitikei Valley (Te Punga 1952; Milne 1973b). Te Punga (1952) recognised seven episodes of terrace formation within the Rangitikei, matching heights of terrace treads on either side of the present day river channel and linking these episodes to changes in relative base level. Milne (1973b) mapped the Rangitikei river terraces in detail, subdividing them into aggradational and degradational and recognising nine major terrace sets (Table 1). Aggradational terraces are associated with glacial and stadial periods during the Quaternary, predicted as a response to increased sediment yields from deforested highlands during cold periods (Milne 1973b; Pillans 1994). The correlation between glacial/stadial periods and river aggradation is based on dating of Ohakean terrace alluvium (Milne 1973b; Marden and Neall 1990; Pillans et al. 1993; Pillans et al. 1996; Vandergoes et al. 2013; Lowe et al. 2015) and the occurrence of subalpine pollen assemblages in Marton terrace alluvium south of Levin (Barnett 1984; Palmer et al. 1988).

Table 1. Major aggradational terraces in Rangitikei Valley. Ages from Milne (1973b), Pillans et al. (1993) and Pillans et al. (1996).

Terrace	Age
Ohakea	10–30 ka
Rata	30–50 ka
Pourewa	70–80 ka
Cliff	90–100 ka
Greatford	110–120 ka
Marton	140–170 ka
Burnand	240–280 ka
Aldworth	340–350 ka
Waituna	360–370 ka

The Rangitikei District is characterised by moderate temperatures without great extremes including warm summers (15–22°C) and mild winters (1–10°C) (Radcliffe 1976; Cowie and Rijkse 1977; Chappell 2015). The climate supports a range of land uses with sheep and beef farming dominating in the Rangitikei River and Pourewa Stream catchments. The threat of drought is present during some summers, particularly towards the coast where it can be drier (Heerdegen and Shepherd 1992). Average temperatures range from a high of around 18°C in February to a low of around 9°C in July (Chappell 2015). On average rainfall does not show a strong seasonal trend; the catchment receives between 1000 and 1100 mm per year. Evapotranspiration follows a seasonal pattern leading to potential soil moisture deficits in summer, depending on the soil type and its ability to retain water (Cowie and Rijkse 1977). Soil moisture deficits occur on average 10–15 days in both January and February and commonly into March and April, with irrigation used to offset the difference (Burgess 1988; Hedley and Yule 2009). The Rangitikei District is exposed to prevailing west to northwest winds that occur 30–50 percent of the time.

The physiography of the Pourewa Valley has been influenced by the same set of processes and forcing mechanisms as the Rangitikei River valley, with the Rangitikei River itself also playing an influential role during landscape development. The Pourewa Stream differs substantially from the Rangitikei River in that its headwaters are located within the soft Plio-Pleistocene sediments of the WB, in contrast to the considerably harder Torlesse greywacke and volcanic rock comprising the headwaters of the Rangitikei River. The Pourewa stream has a shallow gradient, meandering planform with a sandy mud bed until near Hunterville where large quantities of reworked gravel enter the stream from tributaries draining the Rangitikei River terraces (Figure 2). The terraces form extensive cropping and grazing land on the eastern interfluve c. 100 m above the bed of the Pourewa Stream. From this point onward, the Pourewa Stream displays a wandering planform (Neill 1973; Ferguson and Werritty 1983; Fuller 2008) similar to the Rangitikei River albeit on a much smaller scale. Flood control works installed from the 1950s onward have played an important role in constraining the lower Rangitikei River and Pourewa Stream, limiting substantial natural avulsion processes evident from geomorphological mapping (Te Punga 1952; Fleming 1953; Richardson and Fuller 2010; Rees et al. 2018b).

Figure 2. (A) Wandering channel morphology within the lower reach of the Pourewa Stream (NZGD2000 -40.046606 S, 175.456338 E). (B) Influx of Late Quaternary gravels from uplifted Rangitikei terraces capping hills within the central Pourewa catchment 2 km east of Hunterville (NZGD2000 -39.944884 S, 175.581436 E). (C) Confluence between Pourewa Stream and the Rangitikei River displaying a gravel fan, note this fluvial interface undergoes rapid change (NZGD2000 -40.079352 S, 175.452476 E). (D) Pourewa Stream cutting overtop of ‘basement’ composed of carbonaceous siltstone containing the estuarine dwelling cockle Austrovenus stutchburyi, Okehu Group (c. 1.58 – 0.99 Ma) (NZGD2000 -39.953241 S, 175.563025 E). (E) Typical meandering bed form of the Pourewa Stream within its upper reach, note photograph taken following moderate rain, which has led to surface water pooling within paleo-channels across Poukiore Valley (NZGD2000 -39.886163 S, 175.609932 E).

Major tributaries, Ongo Stream and the informally named Leedstown Stream (Figure 1) drain the western side of the Pourewa catchment providing large contributions to the catchment’s otherwise narrow area. The Pourewa Stream sits high in the landscape, flowing SW across its Late Pleistocene–Holocene valley fill. The stream pattern is unusually linear, running approximately parallel to the more deeply incised Rangitikei River for c. 30 km.

The Manawatu plains in the eastern Whanganui Basin contain a series of NE-SW trending anticlines, including the Pohangina, Feilding, Marton, Mt Stewart-Halcombe, and Oroua anticlines (Feldmeyer et al. 1943; Te Punga 1952, 1957). They are bounded on their eastern side by westward dipping reverse faults that only rarely reach the surface (Melhuish et al. 1996; Jackson et al. 1998; Begg et al. 2005). The anticlines are asymmetric, characterised by gentle western slopes (c. 2°) and steep eastern slopes (up to 70°), a geometry interpreted to reflect the underlying reverse faults (Jackson et al. 1998).

Feldmeyer et al. (1943) and Te Punga, (1957) both suggested that these anticlines were formed by ridges of basement rock trending approximately parallel with the main axial range. Structural interpretation of the Whanganui-Manawatu area indicates that these folds are long lived structures that are still actively growing (Jackli, 1957; Lensen, 1976; Te Punga, 1957). Preservation of the 310 ka Brunswick marine terrace tread (Pillans 1983, 1990, 1994) on the anticline crests suggests a majority of the present day topography has formed over the last c. 300 ka.

The Pourewa Stream extends across several different physiographic zones, originating in steep mudstone hill country of the upper Poukiore Valley (Figure 1) to Rata where the stream meanders out over a broad flight of Rangitikei River terraces before finally becoming deeply entrenched as it grades to the Rangitikei River.

Methods

During this study, we compiled a 1:25 000 geological map of the central Rangitikei District, a simplified version of which is displayed in Figure 3, presenting the main geological groups and important regional structures. This map compilation incorporated studies by Feldmeyer et al. (1943), Te Punga (1952), Fleming (1953), Naish and Kamp (1995), Journeaux et al. (1996), Van Der Neut (1996), Potter (1984) and Bussell (1984). We constructed geological cross sections trending east-west and north-south across the study area to aid with structural interpretation (Figure 4). Details of the new 1:25 000 geological map and corresponding geological cross sections of the Central Rangitikei District are described in Rees et al. (2018b). This work was the result of an extensive field campaign from 2016 to 2018.

Figure 3. Simplified geological map of the study area showing the location of the Pourewa Stream catchment outlined in black in relation to major regional structures and lithological groups.

Figure 4. Simplified cross sections A to A’ and B to B’ trending N-S and E-W respectively, displaying the main geological structures of the Rangitikei District. Geological units have been ordered into their respective groups.

During the fieldwork phase of this study, particular focus was placed on observing the movement of water in the landscape and exploring relationships between water flow and the local geology. Springs, seeps and water bearing layers were noted with GPS (Figures 5 and 6) and all available stratigraphic sections logged. Particular lithologies were targeted for sediment sampling and analysis, selected to be representative of key elements within the large suite of sediments encompassed by the mapping area. Samples were taken with a percussion corer to a depth of typically 1–2 m and with stainless steel rings able to be driven into an outcrop by hammer to take smaller 50–150 mm long cores.

Figure 5. (A) Groundwater seeps at the contact between Tuha Sandstone and underlying siltstone of the Tikapu Formation, Rangitikei Group, exposed in cliffs in the Rangitikei River valley (NZGD2000 -39.877205 S, 175.721254 E). (B) Groundwater seeps at a particularly coarse sandstone overtop of a hard well consolidated siltstone forming the ledge in the lower part of the section, Vinegar Hill Formation, Rangitikei Group (NZGD2000 -39.92932 S, 175.642462 E). (C) Groundwater seeps within Mangaonoho Formation, Rangitikei Group at contact with laminated sandstone over massive siltstone (NZGD2000 -39.911645 S, 175.620189 E). (D) Groundwater seeps along the contact between a sandstone and underlying siltstone unit within Rewa Formation, Okehu Group (NZGD2000 -40.008355 S, 175.589676 E).

Figure 6. (A) Small spring flowing out of Holocene gravels at the base of Leedstown Valley (NZGD2000 -40.019826 S, 175.471315 E). (B) Large set of springs along contact between Ratan Gravels and underlying siltstone of Waiomio Formation, Kai Iwi Group on the immediate up thrown side of Rangitira Fault (NZGD2000 -40.023638 S, 175.479813 E). (C) Large set of springs emerging mid to top slope beneath the tread of the Greatford Terrace (MIS 5) (NZD2000 -39.989024 S, 175.511184 E). (D) Medium to small spring emerging from coarse pumiceous sediments overlying fine-grained Coutts Creek Horizon, Okehu Group. Note the typical landform of hummocky landslip topography with water logged soil tolerant species such as rushes and tussock prevalent, typical of a spring site (NZGD2000 -39.940442 S, 175.432777 E).

Due to the wide range of sedimentary deposits within the c. 2000 m of WB strata that the Pourewa Stream cuts through and time constraints on field work, it was necessary to adopt a simplified approach to sediment sampling and analysis. We subdivided the lithologies into broad groups based on grain size and petrography, initially selecting a set of representative sandstone, siltstone, shellbed, pumiceous and conglomerate units. Following initial sediment sampling and analysis, other samples from the main categories outlined above were collected in order to measure some of the variability observed in outcrop. Three samples were also taken from a modern day gravel bed of the Pourewa Stream in three locations moving away from the stream, allowing for collection of samples containing low, moderate and high percentage of fines for later comparison. A single sample from a modern day Rangitikei River gravel bed was also sampled for comparison. The goal of the sampling regime was to provide a preliminary overview of the study areas hydrogeological units by outlining their modal characteristics.

Three 50 × 46 mm sub-samples were taken for triplicate dry bulk density, gravimetric and volumetric water content, total and effective porosity and particle density determinations using methods outlined in Gradwell and Birrell (1979), Claydon (1997) and Rayment and Lyons (2011). Three 150 × 46 mm sub-samples were taken for triplicate measurements of hydraulic conductivity using a falling head permeameter (Figure 7). The falling head permeameter consisted of a 6 m long standpipe with an inside diameter of 50 mm. The standpipe then enters a series of 90° bends before connecting to a sample chamber consisting of a 150 mm length of core with a 46 mm inside diameter. Filter cloth is placed above and below the sample chamber to stop debris falling into the pipe then clamped into place using a combination of hose clamps and rubber sleeves. Samples were first allowed to saturate and reach a steady state of flow before the test was run. Incremental measurements were made of the head in the stand pipe using a water level sounder over time as the head reduced. The timing of measurements was based on the rate of flow for the particular sample; for example coarse grained units required more frequent minute to hourly measurement intervals whilst siltstone and mudstone units were measured twice daily. Hydraulic conductivities were calculated using Darcy’s equation as outlined in Landon et al. (2001).

Figure 7. (A) Photo of the falling head permeameter setup at Massey University, Palmerston North, consisting of a 6 m standpipe tied to the railing within the stair well. (B) Close up of the base where the standpipe enters two 90-degree bends and a main shut off valve before entering the 150 × 4.6 mm sample housed in piping attached by rubber sleeves and hose clamps. The outlet at the top of the sample is epoxied into a plug allowing the outflow of water to enter the 20L water container. Note a second valve on the far right, positioned low to allow the water to be drained from the system following a run.

Three subsamples of typically 100 g were taken for grain size analysis. Grain size analysis was undertaken by dry sieving the >63 micron fraction using a set of sieves at half phi increments. The <63 micron fraction was determined in solution using a Horiba Partica LA-950V2 laser particle analyser, a refractive index of 1.52 and an absorption value of 0.1 was used for the sample material, based on a mixed mineral assemblage of dominantly marine sediments (Rothwell et al. 2006).

Grain mounts were made from the 125–250 μm fraction in each representative lithology and 400 grains were counted to determine the general mineralogy of the sediments. All samples were analysed for major and trace elements by X-ray fluorescence spectroscopy (XRF) at Massey University following procedures and analytical techniques outlined in Kimura and Yamada (1996) and Lee et al. (1997) using a Bruker S8 TIGER with Bruker SPECTRA-plus software. H₂O and loss on ignition (LOI) were calculated from the weighed samples.

The general clay mineralogy of samples was determined from suspension-oriented x-ray diffraction (XRD) samples using the sample preparation techniques outlined in Whitton and Churchman (1987). Measurements were made using a GBC Scientific EMMA diffractometer operating at 35 kV and 20 mA, using monochromatic Co Kα radiation and a scanning rate of 2° 2θ/minute, scanning from 5° to 40° 2θ. Triplicate suspensions for each sample were prepared including non-saturated, K saturated and Mg saturated. Following initial XRD runs, Mg saturated samples were sprayed with glycerol and analysed until complete expansion of smectites was achieved, followed by heating to 550°C and repeat analysis. Resulting XRD patterns were compared with the International Centre for Diffraction Data (ICDD) library to aid with identification of clay minerals.

Total carbon, nitrogen, hydrogen and sulfur concentrations were determined using the elementar® (vario MICRO cube) at Massey University. Inorganic carbon was determined by digestion of samples with hydrochloric acid within a vacuum chamber following the methods outlined by Bundy and Bremner (1972) and Tiessen et al. (1983). Inorganic carbon and total carbon determinations were used to calculate organic carbon content of each sample.

Results

Geology

The geology of the Central Rangitikei District has been mapped at 1:25 000, a simplified version of which is presented in Figure 3 with accompanying structural interpretation (Figure 4). This mapping resulted in the fusion of several formal and informal stratigraphic nomenclatures introduced in the literature over the last 75 years, including work by Feldmeyer et al. (1943), Te Punga (1952), Fleming (1953), Milne (1973b), Seward (1976), Potter (1984), Bussell (1984), Abbott (1992), Naish and Kamp (1995) and Van Der Neut (1996). New regional structures including the Poukiore Fault and a southern extension of the Mangaone Anticline (Feldmeyer et al. 1943) have been identified during the field survey. Major regional reverse faults, the Rangitikei and Rauoterangi faults with maximum-recorded throws of 100 and 200 m west respectively (Naish and Kamp 1995), drive deformation of the Plio-Pleistocene succession, with associated zones of antithetic, normal faulting (Pakihikura and Poukiore faults). In the south of the mapping area the actively growing Marton and Feilding anticlines are observed, related to faulting at depth (Melhuish et al. 1996; Jackson et al. 1998; Begg et al. 2005).

The distribution of the Pouakai to Paparangi group succession (c. 3–0.3 Ma) is related to the depositional history and stratigraphic architecture of the WB through the Plio-Pleistocene. A general east-west alignment reflects the gentle dip (2–10°) of the succession toward the basin depocentre, located to the south. Quaternary terraces represent climatically controlled aggradational episodes preserved by ongoing regional uplift over the last c. 370 ka (Pillans 1994). The terraces fan out in the lower Rangitikei Valley where they dominate the southern portion of the mapping area. Holocene valley fill is restricted to the main valleys where the rivers have incised through the Plio-Pleistocene marine succession.

An important relationship observed in the field is the general increase in level of compaction and cementation with age, reflecting progressively greater depth of burial and level of diagenesis. Mudstones and siltstones typically have a higher level of cohesion and consolidation due to a greater proportion of silt to clay sized particles and tight grain fabric, imbuing them with higher average strength (45–180 KPa) relative to their sandstone counterparts (9–90 KPa) (Ker 1973; Thompson 1982; Ker 1991). For instance, Thompson (1982) reports an average strength determined by point load tests of 156 KPa for Mangaweka Mudstone and conversely 7 to 68 KPa for sandstones of the Rangitikei Group, relating this change to relative levels of consolidation. Conglomerate and gravel units are frequently cemented by iron oxide; typically, these deposits are <1.6 Ma and are dominated by greywacke clasts, with the level of cementation observed to increase with age (Te Punga 1952; Fleming 1953; Milne 1973b). Locally, shell and sandstone beds are cemented by calcite, dramatically increasing their strength (upto 2,857 KPa); however, these deposits form a small proportion of the entire succession (Thompson 1982). Hydraulic conductivity gradually decreases with increasing age (Begg et al. 2005), reflecting an increase in compaction, consolidation and cementation as a function of diagenesis (Houseknecht 1987).

Analyses

Grain size statistics were calculated using the Folk and Ward method (Folk and Ward 1957) and compared with the hydraulic conductivity measurements (Table 2, Figure 8). Increasing hydraulic conductivity is associated with an increase in mean grain size and a reduction in the presence of a fine tail fraction. The samples are typically poorly to very poorly sorted; however, the sample with the highest hydraulic conductivity is moderately well sorted. Overall the hydraulic conductivity measurements range from 5 to 20379 m/day. Coarse grained gravel, shellbed and pumice deposits obtain the highest hydraulic conductivities of between 351 and 20379 m/day, consistent with gravel to granular sediment values documented by Freeze and Cherry (1979), Domenico and Schwartz (1998), Fetter (2001) and New Zealand based studies (White 1986; Thorpe 1992; Davidson and Wilson 2011). Mudstone and siltstone hydraulic conductivities were fairly high at between 5 and 6 m/day, falling between a fine sandstone to silty sandstone range relative to published values (Freeze and Cherry 1979; Fetter 2001). Sandstone units ranged from 18 to 39 m/day, falling within a typical sandstone range when compared with published datasets (Chapuis et al. 1989; Wang et al. 2008).

Figure 8. Graphs demonstrating key relationships between grain size statistics and hydraulic conductivity. (A) Percentage fines content vs hydraulic conductivity (m/day, log scale). (B) Mean grain size (μm, log scale) vs hydraulic conductivity (m/day, log scale).

Table 2. Average grain size, hydraulic conductivity (K), bulk density, total porosity and effective porosity measurements from the representative lithologies of the study area, in order of increasing hydraulic conductivity.

Lithological Unit	Grain size (μm)	Sorting	Hydraulic Conductivity (K) m/day	Bulk Density g/cm3	Total porosity (%)	Effective Porosity (%)
Pourewa Stream gravel (no fines)	Coarse gravel (19102)	Moderately well sorted	20379	1.58	44	43
Waitapu Shell Conglomerate	Very coarse sand (1953)	Very poorly sorted	400	1.67	37	33
Mangamako Shellbed	Very coarse sand (1280)	Very poorly sorted	389	1.37	48	32
Potaka Pumice	Medium sand (316)	Poorly sorted	351	0.89	66	7
Pourewa Stream gravel (moderate fines)	Medium gravel (15924)	Poorly sorted	87	1.75	37	31
Pourewa Stream gravel (High fines)	Very fine gravel (3640)	Very poorly sorted	75	1.70	35	28
Hautawa Shellbed	Fine Sand (125)	Very poorly sorted	51	1.73	35	6
Rangitikei River gravel (High fines)	Medium gravel (9978)	Poorly sorted	48	1.81	31	29
Mangamako Sandstone	Very fine sand (88)	Poorly sorted	39	1.75	34	9
Te Rimu Sandstone	Very fine sand (90)	Poorly sorted	24	1.66	38	7
Rewa Pumice	Very fine sand (67)	Very poorly sorted	18	1.24	48	7
Mangaweka Mudstone	Coarse silt (25)	Poorly sorted	6.2	1.61	39	13
Mangarere Siltstone	Coarse silt (22)	Poorly sorted	5	1.54	42	7

Average bulk densities of the samples range from 1.37 to 1.81 g/cm³ with particularly low values for pumiceous units (Table 2). The sedimentary units measured in this study plotted within similar ranges as unconsolidated Holocene (Avnimelech et al. 2001; Rose and Kuehl 2010) to weakly and moderately lithified Pleistocene sediments (Jakobsson et al. 2001; Bahr et al. 2005; Lamb et al. 2007; O'Regan et al. 2008) in the literature. The particle densities of the stratigraphic units were calculated to be around 2.6 to 2.8 g/cm³, apart from the pumiceous rich sediments, which were calculated to be around 2.4 g/cm³. Total porosities ranged from 34% to 66%, with the highest porosities occurring within very pumiceous sediments and coarse shellbed material. Total porosity values obtained in this study are similar to values presented by Halley and Schmoker (1983) and Boving and Grathwohl (2001) for a comparable range of sediments and sedimentary rocks (Table 2). Effective porosity values range from 43% to 7% (Figure 9), comparable to values presented in Dong et al. (2010) and Medina et al. (2011). The siltstone, mudstone and sandstone samples with relatively high total porosity values of 39%–42%, have much lower effective porosity values of 7%–13%, reflecting the presence of small and intermediate sized pores and therefore, low macro-porosity (Dewhurst and Aplin 1998).

Figure 9. Graphs displaying the relationships between porosity and hydraulic conductivity. (A) Total porosity (%) vs hydraulic conductivity (m/day, log scale). (B) Effective porosity (%) vs hydraulic conductivity (m/day, log scale).

Clay mineralogy of the sediments is dominated by illite with lesser amounts of glauconite, kandite and minor smectite. Detrital glauconite grains were also observed as a minor constituent in the 125–250 μm fraction of 80% of samples during point counts. The mineralogy determined by point counts was dominated by quartz and highly altered minerals including some chlorite, with lesser amounts of feldspar and minor pyroxene and hornblende. Moderate lithic and opaque grains were also present in several coarse grained samples including Mangamako Shellbed, Te Rimu Sandstone, Rangitikei River gravel and Mangamako Sandstone. Weathered clear rhyolitic glass shards dominated the pumiceous deposits while lithic shell material dominated the shellbeds including minor quantities of foraminifera.

Major element chemistry (Table 3) determined by XRF displays changes in geochemistry consistent with the measured suites of minerals. Sandstones show an enrichment in silica and relative depletion in aluminum, iron, and magnesium compared with mudstone and siltstone samples, representing a relative dominance of quartz within sandstone units and higher proportion of altered clay minerals within the siltstone and mudstone. The siltstone and mudstone units also have slightly elevated levels of titanium and potassium compared with the sandstone units suggesting an increase relative to quartz of potassium feldspar (orthoclase and microcline) and minerals such as pyroxene and mica where titanium can occur as an accessory or substituted element (Murray 2006). Calcium values are highest within the shellbed and shell hash conglomerate units at between 7.4% and 30%, due to the presence of abundant detrital shell material within these samples. Correspondingly, these samples display the highest loss on ignition due to carbonate volatilisation during combustion. Samples containing a strong greywacke and/or volcanic provenance namely the shell hash conglomerate, Holocene gravels and pumiceous units display relatively high sodium at between 3.5% and 4.4%, attributed to higher proportions of plagioclase feldspar and vitric rhyolitic glass shards within their respective mineral assemblages.

Table 3. Major element geochemistry (% weight) of respective samples grouped by lithology. Hu – Huatawa Shellbed; Mm – Mangamako Shellbed; Ma – Mangamako Sandstone; Tr – Te Rimu Sandstone; Mw – Mangaweka Mudstone; Mg – Mangarere Siltstone; P – Rewa Pumice; Rg – Rangitikei Gravel; Pg – Pourewa Gravel.

Lithological Group	Sample Code	SiO2	Al2O3	TiO2	MnO	Fe2O3	MgO	CaO	Na2O	K2O	LOI	Sum
Shellbeds	Hu	61.48	12.54	0.52	0.05	3.80	1.35	7.39	2.60	2.13	8.53	100.53
	Mm	35.05	4.47	0.16	0.04	1.44	0.59	29.98	1.21	0.84	25.36	99.30
Sandstones	Ma	76.97	10.57	0.49	0.04	2.64	0.84	1.25	2.38	1.87	2.48	99.60
	Tr	73.62	12.36	0.54	0.05	3.29	1.19	1.41	2.62	2.32	2.28	99.80
Mudstone	Mw	69.98	14.51	0.65	0.04	3.77	1.37	1.04	2.62	2.36	4.03	100.43
Siltstone	Mg	63.42	16.50	0.67	0.07	5.30	2.03	2.09	2.91	3.11	4.29	100.55
Pumiceous sandstone	P	68.35	14.20	0.37	0.05	2.78	0.68	2.05	3.50	2.00	4.77	98.81
Gravels	Rg	70.00	14.63	0.43	0.07	3.52	1.22	2.62	4.39	2.00	1.76	100.75
	Pg	75.84	11.39	0.35	0.04	3.01	0.78	1.47	3.48	1.75	2.31	100.50
Shell hash conglomerate	Wai	58.66	12.38	0.34	0.10	3.92	1.54	11.23	3.57	0.99	8.17	101.00

Carbon (C) determinations show high amounts of inorganic carbon within the shellbed and shell hash conglomerate units, Mangamako Shellbed being the highest outlier at 60.7 g of C per kg (Table 4). The majority of other samples display inorganic carbon values >2 g of C per kg. Pumiceous and shellbed sediments also display the highest organic carbon contents of up to 8.6 g per kg. Sandstones contain less organic carbon at 0.5 to 0.6 g per kg, compared to mudstones and siltstones with higher values of 0.9 to 1.6 g per kg.

Table 4. Average organic and inorganic carbon contents and clay minerals identified during analysis of respective lithologies, in order of increasing organic carbon content.

Lithological Unit	Organic carbon (g of C per kg of sediment)	Inorganic carbon (g of C per Kg of sediment)	Clay minerals present (XRD)	Glauconite as percentage of 125-250 μm fraction.
Mangamako Shellbed	8.60	60.69	Illite, Kandite	2.0%
Rewa Pumice	8.46	0.43	Glauconite, Illite	0.0%
Waitapu Shell Conglomerate	3.30	19.29	-	1.5%
Hautawa Shellbed	2.67	10.52	Illite, Kandite, Smectite	2.2%
Pourewa Stream gravel	2.15	1.94	Glauconite, Illite, Kandite	1.5%
Mangaweka Mudstone	1.59	0.30	Glauconite, Illite	0.0%
Mangarere Siltstone	0.88	1.61	Illite, Kandite	0.2%
Mangamako Sandstone	0.63	0.26	Glauconite, Illite, Kandite	0.8%
Te Rimu Sandstone	0.52	0.18	Glauconite, Illite, Kandite	0.5%
Rangitikei River gravel	0.37	1.93	Glauconite, Illite, Kandite	1.5%

Note we were unable to identify any clay minerals within Waitapu Shell Conglomerate due to poor diffraction patterns.

Hydrostratigraphy

Figure 10 displays a map of potential aquifers in the central Rangitikei District, separating different geological units into hydrostratigraphic groups based on their age and dominant lithology. Aquifer and aquitard units were determined by assigning average hydraulic conductivity, porosity, grain size and bulk density measurements to each geological unit. Given the sampling regime used in this study It was necessary to assign values derived from a units’ dominant lithology. For instance, Tikapu Formation, Rangitikei Group (Naish and Kamp 1995) contains shellbed and siltstone members yet was assigned average properties from its dominant lithology siltstone, resulting in an aquitard classification. Figure 10 also plots the springs observed during the fieldwork in the study area. The spring data represents only a small portion of total springs and seeps in the study area. The grouping of sedimentary units into aquifer and aquitard layers is very broad and serves only to show general regional trends as opposed to detailed local scale aquifer spatial patterns.

Figure 10. Simplified hydrostratigraphic map of the study area showing the location of the Pourewa Stream catchment outlined in black in relation to the local geology. Geological units subdivided based on average grain size, hydraulic properties, age and depositional regime into respective hydrostratigraphic units. Major regional structures are also shown alongside mapped spring locations; note the mapped springs only present a small proportion of the total springs in the area, a majority of which are minor seeps and small springs.

Discussion

Landscape evolution

The widespread aggradational terraces of the Rangitikei are prominent features in the present day physiography of the Pourewa Stream catchment (Figures 3 and 11). Exceptionally thick and widespread Ratan terraces (30–50 ka) occur near Rata (Figure 1) where the Pourewa Stream emerges onto the open, lower Rangitikei River valley. These gravels are derived from the paleo-Rangitikei River and pre-50 ka Rangitikei valley fills, filling areas of the Pourewa valley to depths of up to 10 m. The confluence of Leedstown Stream and Pourewa Stream is interpreted to have been subjected to major scouring post 50 ka given the occurrence of particularly thick Ratan gravels overlain by rare dunes formed by wind-blown silt off an aggrading fluvial floodplain during the last glacial maximum (LGM) (Milne 1973a; Palmer and Pillans 1996). These local areas of fluvial scour and fill are observed in the present study to host important shallow ground water reserves, related to large freshwater springs.

The course of the Rangitikei River has obviously been located much further west of its present position within the structural low of the Pakihikura Fault Zone. A series of progressively younger terrace treads stepping down from the land immediately north of Marton to the present day flood plain of the Rangitikei River suggest the river has tracked eastward over the last 370 ka, likely in response to offset along major NE-SW trending faults and increasing confinement by the growing Marton and Mt Stewart-Halcombe anticlines (Figure 3). Fifty thousand years ago, it is likely the Pourewa stream was substantially shorter than present day, exiting its valley carved through the Plio-Pleistocene rocks of the WB at Rata into the paleo-Rangitikei River (Figure 11).

Figure 11. Schematic palinspastic diagram of the lower Pourewa Stream displaying the relatively rapid landscape changes that have occurred over the last 50 ka. Outlines of the modern day Rangitikei River terrace treads are displayed representing the maximum preserved extent of the Rangitikei River for each approximate period of time. These lines indicate the boundary between the terrace tread (age annotated) and the terrace riser of the next (older) terrace. In the case of the Aldworth line, this indicates the maximum preserved extent of the tread before it gives way to hill country. The modern day confluence between the Rangitikei River and Pourewa Stream is indicated by a red star illustrating the amount of eastward migration the Rangitikei River has undergone in the last 50 ka contemporaneous with development of the lower Pourewa Stream. Data sourced from Te Punga (1952), Milne (1973b), Hellstrom (1993), Pillans (1994) and this study.

Since 50 ka ongoing uplift manifested as offset along regional faults has led to further eastward migration of the Rangitikei River through the LGM aggradation and multiple Holocene degradation episodes. Contemporaneous fluvial development has resulted in a fault controlled extension of the Pourewa Stream to the SW, extending from Rata parallel to the Rangitira Fault. We infer the Pourewa Stream has preferentially exploited this fault zone as the path of least resistance during its natural inclination of cutting down to meet the Rangitikei River. Another line of thought is that the Rangitikei River itself could have produced the substantial fluvial scouring at the confluence of the modern day Pourewa and Leedstown streams, so that the Pourewa Stream is exploiting a topographic low carved into the Pleistocene marine sediments by a paleo-channel of the Rangitikei River.

The Poukiore Valley (Figure 1) at the headwaters of the Pourewa Stream has been influenced by a broad structure in the underlying Plio-Pleistocene sediments. The Mangaone Anticline (Figure 3) forms the western watershed boundary of the Pourewa Stream, defining a marked change in dip direction within the Plio-Pleistocene sediments. Sediments on the eastern side of the structure dip to the SE while equivalent sediments to the west dip W to SWW into a structural low occupied by the Turakina River. Sediments along the western watershed of the Pourewa Stream typically dip to the SE and as a result springs and seeps outcrop extensively on the southern and eastern slopes of this hill country, some providing domestic drinking water.

Hydrostratigraphic map

The hydrostratigraphic map (Figure 10) displays two key relationships that highlight important factors influencing the hydraulic properties of the areas lithologies. Firstly, the age of the sedimentary units is a controlling factor, determining the amount of diagenesis and resulting cementation and compaction a unit has undergone (Larsen and Chilingar 1983; Bjorlykke 2014). Holocene fluvial deposits formed over the last c. 11 ka (Holocene gravels and sands on Figure 10) represent the least consolidated units in the mapping area with little to no cementation and low compaction, forming the most productive shallow unconfined aquifer within the study area (Begg et al. 2005). In contrast, Te Rimu Sandstone (Pliocene sandstone on Figure 10), including the basal sandy Mangarere Formation (Naish and Kamp 1995) represents a succession of locally cemented coarse to fine siliciclastic sandstone that has undergone diagenesis since deposition c. 2.58–2.8 Ma (Wilson and Roberts 1999). Because of a relatively high level of compaction and local cementation, this unit contains substantially lower hydraulic conductivity than contemporary Holocene gravelly sands, illustrated by a change from 24 to 75 m/day respectively (Table 2).

The second key relationship observed from Figure 10 is the difference in depositional setting within the hydrostratigraphic units. Clearly, the Late Quaternary and Holocene gravels and sands fill the modern day drainage network of the Rangitikei Valley, formed over the last 370 ka, throughout a series of climatically controlled aggradational and degradational episodes (Milne 1973b; Pillans 1994). The abundance of greywacke and andesite gravel within these deposits imparts high hydraulic conductivity.

The increasingly common occurrence of greywacke gravel within the Late Quaternary hydrostratigraphic units of the study area reflects substantial changes in the paleogeography and sediment transport pathways of the lower North Island (Te Punga 1952; Fleming 1953; Beu et al. 1981; Melhuish et al. 1996; Rees et al. 2018a). Notably, the onset of volcanic activity within the Tongairiro Volcanic Centre (c. 350 ka) (Shane et al. 2017) providing a new provenance for volcaniclastic sediment, delivered by rivers draining the central volcanic plateau. Simultaneous uplift along the main axial range front (Beanland 1995; Jiao et al. 2015) has provided a source of abundant greywacke detritus, as the modern day topography formed. Localised uplift along a series of regional anticlines (Te Punga 1957) has acted to shape and constrain the emerging fluvial drainage network, funnelling a series of high permeability sediments between bounding basement highs.

In contrast, the linear nature of the Plio-Pleistocene marine sediments exposed along depositional strike, record the paleogeography and depositional setting at their time of formation, characterised by shallow marine deposition within a southward migrating marine depocentre adjacent to the uplifting paleo-axial range (Anderton 1981; Rees et al. 2018b). A key difference distinguishing this older marine basin fill from the overlying Late Quaternary sediments, is the occurrence of relatively thick (c. 10–60 m) (Abbott 1994), massive, blue grey, siltstone units representing high stand deposition within inner to mid shelf depths (Carter and Naish 1998). We suggest these units attain enough lateral persistence to act as local confining layers and as such form an important constituent of the hydrostratigraphy.

Physical and hydrogeochemical attributes

The most important factors related to a sample’s hydraulic conductivity were the presence of a fine tail fraction followed closely by mean grain size. This correlation is consistent with known relationships between grain size and hydraulic conductivity determined by experimental evidence (Krumbein 1934; Hazen 1983; Alyamani and Sen 1993; Song et al. 2009; Sen 2015) and field studies (Fogg et al. 1998; Odong 2008; Vienken and Dietrich 2011). The general reduction of hydraulic conductivity with a corresponding increase in fine content (Figure 8) has been well documented by Schalchli (1992), Benson and Trast (1995) and Boadu (2000). For example, the three gravel samples taken from Pourewa Stream show an increase in hydraulic conductivity from 75 to 20379 m/day, corresponding to a decrease in fines content from high to no fines (Table 2). This marked change occurs despite the gravel component (size, shape) being similar within all three samples.

While the mean grain size is important, it does not show a 1:1 relationship with hydraulic conductivity. For instance, the Pourewa Stream gravel samples with moderate to high fine content have a larger mean grain size (3640 to 15924 μm) yet have lower hydraulic conductivity than Potaka Pumice (mean grain size 316 μm) (Table 2). This could be attributed to the difference in the amount of fines which act to fill voids between the pore spaces, affecting the hydraulic conductivity (Koltermann and Gorelick 1995). Almost all of the samples displayed poor to very poor sorting; however, the sample with the highest hydraulic conductivity (i.e. Pourewa Stream gravels with no fines) was moderately well sorted, demonstrating that sorting is likely another important contributing factor. A wider sampling coverage and increased sampling rate of the sedimentary units within the study area would likely highlight this third key factor.

The siltstone and mudstone units tested in this study exhibited relatively high hydraulic conductivity (5 to 6 m/day) relative to examples from the literature (Freeze and Cherry 1979; Dewhurst and Alpin 1999; Fetter 2001) attaining a closer fit to sandy mudstone and siltstone mixtures documented by Revil and Cathles (1999) and Belcher et al. (2001). This is attributed to a significant component of sand within the mudstone tested, a coarse silt grain size for tested siltstone lithologies and the generally weakly lithified and cemented nature of the WB sediments (Thompson 1982; Ker 1991; Kamp et al. 2004). It is also possible an undetermined amount of bypass flow occurred during permeameter tests, potentially affecting the results and leading to unusually high hydraulic conductivities within the fine-grained sediments measured; however, we consider this error be negligible in our analyses. Methods to combat bypass flow included allowing significant time for the fine sediments to saturate and swell against the sample chamber and the installation of epoxy around the outside of the lower filter cloth covering the bottom of the sample chamber to channel flow toward the centre of the chamber.

Bulk density values typically range between 1.6 and 1.8 g/cm³. Particularly low bulk density values are associated with shellbed and pumiceous units, which correspondingly display the highest total porosity of between 48% and 66%. Generally, total porosity values are high (>30%) and display a weak relationship with hydraulic conductivity indicated by a slight trend of increasing hydraulic conductivity with increasing total porosity (Figure 9). Units containing high total fines e.g. Mangarere Siltstone exhibit high total porosity (42%), a large portion of which is made up of small to intermediate pores with low interconnected macro-porosity (Stephens et al. 1998).

Effective porosity displays a stronger positive correlation with hydraulic conductivity (Figure 9). However, Potaka Pumice, a granular, pure, pumice with high hydraulic conductivity displays unusually low effective porosity. It has proven difficult to determine the correct effective porosity of highly pumiceous sediment during this study. In the laboratory, pumiceous units demonstrate very high water holding capacity, resulting in low calculated macroporosity and therefore, effective porosity. However, from our field experience, pumiceous sediments and soils are very well drained and prone to drought, rapidly drying in the summer and demonstrating low water holding capacity. The effective porosity values of Potaka Pumice are considered erroneous in this study, due to the difficulties encountered in attaining an accurate, representative measurement in the lab.

Another strange outlier (Figure 9 B) is Mangaweka Mudstone, displaying a higher effective porosity (13%) than the two sandstone samples; Te Rimu (7%) and Mangamako sandstone (9%). This discrepancy is likely the result of a relatively high sand content within the Mangaweka Mudstone sample, classified as a sandy mudstone during grain size analysis. Potential error within the effective porosity results may also have contributed to this difference. Hautawa Shellbed displays an unusually low effective porosity (6%) considering its relatively high hydraulic conductivity (51 m/day), considered to be a result of the fine-grained matrix including illite, kandite and smectite clays that provide small to intermediate sized pores and result in high water holding capacity.

Pumiceous units displayed lower bulk density, particle density and higher total porosity than any other sediment type analysed in this study. They also contained low inorganic carbon, but relatively high amounts of organic carbon. This is interpreted to reflect the nature of their depositional environment, typically characterised by rapid fluvial aggradation resulting in potential for organic matter preservation (Gastaldo and Demko 2010), clearly illustrated by the abundance of lignite, charcoal and logs exposed within the volcaniclastic units of the WB (Te Punga 1952; Fleming 1953; Shane 1991; Rees et al. 2018b). Very rich pumiceous units in the WB are the result of episodic ignimbrite emplacement in the central North Island (Shane 1991), rapid erosion of the distal unwelded portion of an ignimbrite sheet results in paleo-fluvial systems becoming chocked with volcaniclastic debris (Manville et al. 2009a; Manville et al. 2009b). Drastically increased sedimentation rates following ignimbrite emplacement (von Lichtan et al. 2016) are interpreted to have led to coastal progradation (Lewis 2007) and the formation of extensive coastal plain and marginal marine deposits in the WB succession (Te Punga 1952; Abbott 1994; Shane 1994; Pillans et al. 2005; Rees et al. 2018b). These deposits are often found in close relation with lignite’s, carbonaceous mudstone and stunted Austrovenus stutchburyi shellbeds in the Rangitikei District, representing high rates of sedimentation within marginal marine to coastal plain environments (Te Punga 1952; Rees et al. 2018b).

The depositional environments of the pumiceous units in the WB imbues them with an inherent lateral variability, which we suggest here to be mirrored by associated variations in hydraulic conductivity of between 18 and 351 m/day. Pockets of high porosity clean pumice such as Potaka Pumice are able to store and transmit significant amounts of water; however, it is extremely difficult to predict their lateral extent from outcrop given rapid facies changes. Groundwater from Maxwell Group sediments in which several volcanic/volcaniclastic units occur together with associated lignite and carbonaceous mudstone are used locally for water supply; however, the water is often stained and considered unpotable (Fleming 1953).

The relationship between a fluvial depositional regime and rapid lateral variations in hydraulic conductivity is also relevant to the Late Quaternary terrace and modern floodplain hydrostratigraphic units, which demonstrate a similar variation in hydraulic conductivity including a change of between 48 and 20379 m/day (Pourewa and Rangitikei gravel samples, Table 2). Seams of well-washed gravels are interpreted to occur along both paleo and modern fluvial channels forming pockets of highly permeable sediment with high hydraulic conductivity surrounded by overbank and alluvial flood plain deposits of mixed grain size with considerably lower hydraulic conductivity, despite considerable gravel content.

Water preferentially flows along the channels of modern day rivers largely as a function of gravity and topography (Helton et al. 2014; Blume and van Meerveld 2015), moving through the landscape eroding, sorting and re-depositing sediment (Kondolf and Piegay 2003). As a result, braided to wandering rivers with highly indurated source rock catchments, such as those occurring across the Canterbury Plains and draining the North Island main axial range contain riverbeds lined with highly permeable gravels, allowing streams to retreat into their beds during periods of low flow (McHugh et al. 2015). We observe this phenomenon in the gravel bedded streams of the Rangitikei District including the Pourewa Stream, through the summer months.

Denitrification

Nitrogen loss from farms due to rapid leaching of nitrate through the soil profile into the subsurface is an increasingly important environmental issue in rural New Zealand, due to increased use of nitrogen fertiliser and land use intensification (Houlbrooke et al. 2004; Parfitt et al. 2006). Subsurface denitrification can result in attenuation of nitrate leaching from agricultural lands (Elwan et al. 2015; Rivas et al. 2015; Collins et al. 2017; Rivas et al. 2017). Natural attenuation of excess nitrate leached from farmland has flow on effects for aquatic systems, reducing water quality impacts and promoting healthy ecosystem functions (Quinn et al. 2009).

The clay mineralogy reported from stratigraphic units in this study are similar to modern day sediments described on the Whanganui shelf (Hume and Nelson 1986; Gillespie 1992; Gillespie and Nelson 1996). Illite imparts natural attenuation within groundwater systems facilitating the adsorption and oxidation/reduction of heavy metal contaminants such as arsenic, cadmium and chromium (Palmer and Puls 1994; Wang and Mulligan 2006; Longe and Enekwechi 2007).

The common occurrence of illite, glauconite and organic matter within the WB sedimentary units and the reworked modern day fluvial deposits of the Pourewa Stream provides evidence for potential electron donors in the subsurface environment (Puckett 2004; Cundy et al. 2008; Debernardi et al. 2008; Rissmann 2011). Glauconite and illite contain structural and exchangeable ferrous iron (Fe²⁺) that can facilitate reduction of nitrate (NO₃^-) in the subsurface to N₂O and then N₂ gas during the microbially mediated process of denitrification (Korom 1992; Ernstsen 1996). Organic carbon within the pumiceous (8 g C per kg) and shellbed (8.6 g C per kg) members of the WB succession could also act as a source of electron donors, supporting the denitrification potential of local aquifers. Even the lower quantities of organic carbon measured within the finer grained siltstones and mudstones (0.8–1.6 g C per kg) could be important if long ground water residence times (years) along flow paths allow sufficient time for redox reactions to occur, leading to reduction of nitrate in the subsurface environment (Seitzinger et al. 2006). Organic carbon preserved within the shellbeds of the WB is interpreted here to be a result of bypassing of siliciclastic sediment on the areas of sea floor allowing for stratigraphic condensation and preservation of moderately low amounts of organic matter.

Inherent limitations incorporated due to the sparseness of sampling within a texturally complex marine basin and modern day fluvial basin must be acknowledged in this study. Continued research is underway into the groundwater systems of the Rangitikei District, aiming to develop an understanding of the role rock type plays in nitrogen attenuation in the subsurface. This work involves sampling of a wider array of sediments within the WB through the application of rapid in situ testing and application of the ideas and concepts developed during this work at a wider regional scale.

Springs

Local residents of the Rangitikei District have made observations of changing rates in spring flow corresponding to changes in atmospheric pressure, with a distinct increase in spring flow occurring during incoming low pressure systems, typically preceding a rainstorm event. Some springs we observed show a visibly higher flow rate following a period of heavy rainfall while others appear to flow at the same rate despite heavy rain, maintaining a constant flow through winter and summer. These two sets of observations lead us to believe there is some evidence for confined aquifer springs within the study area. Unconfined aquifer springs typically are not immediately affected by changes in atmospheric pressure due to a delay in transmitting atmospheric pressure through the vadose zone (Rasmussen and Crawford 1997; Spane 2002); their flow rates are also more likely to be affected by seasonal variations in rainfall due to generally faster recharge times and a higher level of connectivity with surface water (Scanlon et al. 2002). Springs being immediately affected by atmospheric pressure and flowing at similar rates through winter and summer provide anecdotal evidence of confined aquifer flow.

Springs have been related to paleo-channels within the Late Quaternary gravels intercepted by a marine cliff in the Horowhenua District (Kampman and Caldwell 1985; Reynolds 1992; Cussins 1994). Here, we document similar relationships in the Rangitikei District where several springs observed on the Ohakean terraces are related to paleo-channels within the terrace gravels (Figure 12). These channels are only obvious on the LGM aggradational surfaces and younger degradational terraces, due to a thick blanketing of cover beds which mask the micro-topography of pre LGM aggradation terraces (Milne 1973a, 1973b). We interpret the channels to be composed of coarse gravels with high hydraulic conductivity effectively funnelling the shallow groundwater, which flows along the path of least resistance in the landscape. Underlying Pleistocene siltstone forms a less permeable base promoting horizontal groundwater flow on the interface between the gravel and siltstone units. Where terrace risers or cliffs cut by modern day streams intercept the paleo-channels, springs emerge at the contact between the gravel and underlying siltstone (Figure. 12). These springs are typically utlised by farmers who dam one end of the receiving tributary creating a pond, which can then be used to supply water to animal toughs; or alternatively dig out the head of the spring and line the hole to create a small reservoir, which can then be piped by gravity to a tank.

Figure 12. A) Aerial photograph of paleo-channels observable as micro-topography on the Last Glacial Maximum surface (LGM) (NZGD2000 -40.006514 S, 175.576372 E). These channels are related to shallow groundwater springs and seeps where interception occurs by the modern day fluvial erosion of the Rangitikei River and its tributaries. Underlying Pleistocene siltstone dipping gently to the SW creates a sloping aquitard surface beneath the more permeable Late Quaternary gravels. B) Example of the undulating surface of erosion beneath an Ohakean Terrace (LGM surface) (NZGD2000 -40.081486 S, 175.466938 E); fine to medium grained fluvial sand is intermixed with gravel within channel cut and fill overlain by further gravels. This photo is included to illustrate some of the variability within Late Quaternary Terraces. Terrace deposits have been observed with incorporated logs and boulders highlighting the inherent danger of overgeneralising these highly variable hydrostratigraphic units.

Conclusion

The Pourewa catchment lies within the WB and its geology is dominated by alternating coarse and fine-grained sedimentary units deposited during cyclic eustatic sea level fluctuations spanning the entire Quaternary Period. Regional fault zones and related basement highs trending approximately NE-SW exert control on the structure of the Plio-Pleistocene sedimentary units and the development of the modern day fluvial system in the Rangitikei and Pourewa catchments.

Climatically controlled Late Quaternary aggradational episodes have played an important role in the formation of the present day physiography of the Pourewa Stream catchment. At c. 50 ka the Pourewa Stream was substantially shorter than at present day, entering the paleo-Rangitikei River near Rata. Since c. 50 ka ongoing uplift manifested as offset along regional fault systems has led to eastward migration of the Rangitikei River. Contemporaneous fluvial development has resulted in a fault controlled extension of the Pourewa Stream before its eventual decent to Onepuhi and confluence with the Rangitikei River.

Hydraulic conductivity of the main representative lithologies appears to be governed primarily by presence of a fine tail fraction, mean grain size, effective porosity, sorting and level of cementation and compaction as a function of diagenesis and age. Rapid lateral facies variations in the Pleistocene succession are interpreted to be reflected by corresponding changes in hydraulic conductivity, porosity and geochemistry.

Clay mineralogy of the sediments is dominated by illite with lesser amounts of glauconite, kandite and minor smectite, similar to modern day sediments on the Whanganui shelf. Geochemical data display changes consistent with the observed mineral assemblages. Organic carbon is highest within pumiceous and shellbed deposits, interpreted to reflect environments of deposition. Common glauconite, illite and moderately low organic carbon are present within the hydrostratigraphic succession facilitating microbially mediated denitrification in the subsurface through provision of electrons. Ongoing research into these subsurface processes aims to capture a wider proportion of the textural variability in the WB, applying the ideas and concepts developed herein at a regional scale.

Some spring flows in the Rangitikei District have been noted to change with season and atmospheric pressure while others remain relatively constant. We interpret this as evidence for confined aquifers within the study area. Several springs observed on LGM terraces are related to paleo-channels on the overlying terrace treads. We suggest paleo-channels contain coarse gravels with high hydraulic conductivity, funnelling the shallow groundwater, which flows along the path of least resistance in the landscape.

Acknowledgments

We thank the residents in the Central Rangitikei District for granting land access. Special thanks to Massey University staff and laboratory technicians including Ian Furkert, Glenys Wallace, Bob Toes, Ross Wallace, David Feek, Dr Roberto Calvelo-Pereira, Dr Qinhua Shen, Dr Anja Mobis, Dr Gabor Kereszturi and Dr Mike Bretherton. This manuscript has been improved tremendously by helpful reviews from Dr Martin Brook and an anonymous referee.

This paper uses the spelling ‘Whanganui’ rather than ‘Wanganui’ as recommended by the New Zealand Geographic Board (Ngā Pou Taunaha o Aotearoa). The spelling of ‘Pourewa’ rather than ‘Porewa’ is used as recommended by Ngāti Hauiti.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by Horizons Regional Council: [grant Number Don Linklater Bursary Research Grant]; Institute of Agriculture and Environment, Massey University [grant Number C V Fife Memorial Scholarship, Helen E Akers PhD Scholarship, Massey University Doctorate Scholarship].