Biomass and root attributes of eight of New Zealand’s most common indigenous evergreen conifer and broadleaved forest species during the first 5 years of establishment

Study site

The trial site was located on a low-lying, even-surfaced alluvial terrace adjacent to the Taraheru River, in Gisborne City (Fig. 1). The same site has also been used to measure ‘plant growth performance’ of: (i) 12 early (colonising) indigenous species considered typical of riparian margins (Marden et al. 2005); (ii) different clones of poplar and willow (Phillips et al. 2014); and (iii) a range of exotic forest species (Phillips et al. 2015). Temperatures over summer average 23 °C and over winter 12 °C, and mean annual rainfall is 1000 mm. The soil is a naturally fertile, free-draining Typic Sandy Brown Soil of the Te Hapara soil series (Hewitt 2010). Although not used, a drip-irrigation was installed to ensure survival of the plants in the event that drought could jeopardise the longer-term aims of this trial.

The soil has no physical or chemical impediments to root development to about 1.2 m depth, other than a variable-depth water table that fluctuates between > 1.5 m depth and within ~ 0.2 m of the surface (Phillips et al. 2014).

Trial design

The site (50 × 20 m) was tilled, and weed mat was laid down before planting in autumn 2006 (Fig. 1). Weed mat reduced weed competition and minimised efforts to extricate intertwined roots belonging to weeds from the anticipated dense and fibrous root network of the trialled species.

Containerised seedlings sourced from a local nursery consisted of 1.2-L polythene bags (0.12 m deep by 0.10 m diameter) filled with a 50:50 mix of bark and pumice. To minimise root disturbance, seedlings were planted with the soil attached. Root binding was minimal and root pruning before planting was not required as tap root length did not exceed the depth of the planting bag. The site was subdivided into three blocks and all three blocks were planted in 1 day. A randomised block design was used with equal numbers of each species in each block. Trees within blocks 1 and 2 (plants to be extracted 1 and 2 years after planting, i.e. at age 3 and 4 years old, respectively) were planted at 1-m spacing, and in block three (plants to be extracted 3 years after planting, i.e. at age 5) at 1.5–2.0-m spacing to allow for increases in canopy and root growth during the course of the trial.

Species selection

Data collection

A group of 1- and 2-year-old plants were removed from their containers and destructively partitioned into their component parts. At the trial site, another group of 2-year-old seedlings were planted and a subset of plants was excavated and similarly destructively partitioned on the anniversary of planting for three successive years. An air lance at 240 kPa was used to remove the soil surrounding each root system without damage or significant loss of the fine root mass. The aim was to sample ten plants per species per year, but frost killed some puriri plants and stem breakage reduced the sample size of titoki plants (Table 1).

Tree height, canopy spread, root collar diameter (RCD—measured overbark at ground level) and diameter at breast height (DBH—measured at 1.4 m above-ground level) (where applicable) were measured before removing the plants from the trial site (Additional file 2). Canopy spread was taken as the mean of the diameter measured in two directions at right angles. Trees were cut at ground level and the above-ground components separated into branches, foliage and stem (Additional file 3). Root system extraction and measurement methods followed well-established procedures (e.g. Watson et al. 1999; Marden et al. 2005; Phillips et al. 2014, 2015). Once removed from the ground, the root system of each plant was washed to remove adhering soil matter then photographed (an example is shown in Fig. 2). It was then described using the classification of Hinds and Reid (1957) before being partitioned into root bole (stump) and roots. Below-ground growth parameters included mean root depth and mean root spread of the lateral roots, hereafter referred to as root depth and root spread (Additional file 4). The latter was taken as the mean of the maximum root length measured from root tip to root tip in two directions at right angles to each other. Root length (excluding fibrous roots) within each of the root diameter size classes was measured by laying pieces of root end for end alongside a tape measure (Watson et al. 1999). The mass of the roots and root bole were calculated separately (Additional file 5). Root distribution by mass (Additional file 6) and length (Additional file 7) relative to the root bole was obtained by systematically dissecting the root systems into 0.5-m radial and depth segments then sorted by diameter size classes, (< 1 mm fibrous, 1–2, 2.1–5.0, 5.1–10.0 and 10.1–20.0 mm over bark) (Watson and O’Loughlin 1990).

All plant components, both above- and below-ground, were oven-dried at 80 °C for 24 h or until no further weight loss was detectable then weighed to the nearest 0.1 g. The biomass and root:shoot ratio were calculated using dry mass. A carbon content of 50% of biomass is considered robust (Coomes et al. 2002). A conversion factor of biomass × 1.65 (Williams 1978) was used to calculate carbon stocks (t CO₂). Carbon stock values do not include woody debris or fine litter.

Statistical analyses

ANOVA was used to determine the effects of year and species on above- and below-ground parameters. The Student-Newman-Keuls post hoc analysis was used to determine differences among the species within a year and among years when all species were averaged within each year. The normality of each analysis was determined by visual assessment of residual plots. The root biomass and root length data at 0.5-m radial intervals exhibited non-normality due largely to the prevalence of radial segments in which no or little root material was recorded for some of the root diameter size classes. The radially segmented root biomass and length data were therefore transformed using the equation log_e (x + c), where x was the root biomass or root length and c was 1.

Depending on the method of root extraction and focus of the research, regressions of total root biomass and length are often based on different groupings of root diameter size classes. For comparative New Zealand studies, some researchers have based their regressions on roots with a > 1-mm diameter (Phillips et al. 2015) while others have included only roots with a > 2-mm diameter (Watson and Tombleson 2002, 2004; Marden et al. 2016). In this study, when regressed against RCD, a two-tailed, unpaired t test found no significant difference in the slopes of the curves for roots with > 1- and > 2-mm diameter (P = 0.741). Therefore, regressions based on roots > 1-mm diameter were used for analyses of total root length and total root biomass. Unless otherwise stated, data are presented as a mean with standard errors, and statistical analyses were considered significant if P ≤ 0.05.

Differences between the grouped conifers versus broadleaved species were not examined because they differ significantly in tree form; there were clear overlaps in above- and below-ground metrics between species across these groupings and because of the likely bias in analyses resulting from too few and/or uneven sample numbers.

Allometric relationships

Tree height

Linear regression fitted the RCD and tree height data well, with r² ranging between 0.768 (Vitex lucens) and 0.916 (Dacrydium cupressinum), and all regressions were highly significant (P < 0.001; Fig. 3). There were significant differences among species with respect to height in each year (Additional file 2). With the exception of Prumnopitus taxifolia, which had the same average height in years 1 and 2 (P = 0.05: data not shown), the tree height of the remainder of species increased significantly from year to year. There was, however, no clear pattern as to which species had performed best until year 5 when Prumnopitys taxifolia (1.3 m), Agathis australis (1.6 m) and Prumnopitys ferruginea (1.6 m) were significantly shorter (P = 0.05) than the tallest two species Podocarpus totara (2.8 m) and Dacrycarpus dacrydioides (2.7 m), while Vitex lucens (2.4 m), Dacrydium cupressinum (2.2 m) and Alectryon excelsus (2.2 m) reached an intermediary height significantly different from the other species but not each other (Additional file 2).

Above-ground biomass (AGB)

Two-parameter exponential regression analysis was a good fit for the RCD and total AGB data, with r² ranging between 0.803 (Agathis australis) and 0.944 (Dacrycarpus dacrydioides), and all regressions were highly significant (P < 0.001) (Fig. 4). There was, however, no significant difference in mean total AGB among species until years 4 and 5, the exception being Alectryon excelsus which showed no further increase in mean total AGB (P = 0.05; data not shown). By year 5, the mean total AGB amassed by Agathis australis (251.1 g) and Prumnopitys ferruginea (381.4 g) were not significantly different from each other nor from Prumnopitys taxifolia (581.0 g). However, their biomass was significantly less (P = 0.05) than that of Alectryon excelsus (1092.3 g), Dacrydium cupressinum (1165.2 g) and Dacrycarpus dacrydioides (1231.0 g) and was exceeded only by Podocarpus totara (2142.9 g) and Vitex lucens (3026.0 g) (Fig. 5, Additional file 3).

Below-ground biomass (BGB)

Similarly, two-parameter exponential regression analysis was a good fit for the RCD and mean total BGB data, with r² ranging between 0.757 (Podocarpus totara) and 0.946 (Dacrydium cupressinum), and all regressions were highly significant (P < 0.001) (Fig. 6). Additionally, there were significant differences (P = 0.05) between species in the allocation of biomass between root mass and root bole in each year of the trial (Additional file 5). The differences in mean total BGB among species in 1- and 2-year-old seedlings were initially small, but it became more significant in years 4 and 5 (P = 0.05; data not shown), when the majority of species showed a significant increase in mean total BGB. Exceptions to this general pattern include Agathis australis, with no increase in mean total BGB in years 1 and 2, a modest increase in year 3 and no increase in year 4, and in year 5, the increase in mean total BGB was significantly greater than in any of the previous years. By year 5, Agathis australis (72.0 g) had a significantly less (P = 0.05) mean total BGB than both Prumnopitys ferruginea (134.9 g) and Prumnopitys taxifolia (134.6 g). For the remaining species, the mean total BGB of Dacrycarpus dacrydioides (371.2 g), Dacrydium cupressinum (397.4 g), Alectryon excelsus (402.2 g) and Podocarpus totara (450.0 g) was significantly different again but not to each other. Vitex lucens (1257.4 g) had the highest mean total BGB and was significantly different (P = 0.05) from all the other species (Fig. 5, Additional file 3).

Root biomass distribution by depth and radius

For all species in each year of the trial, 100% of the total BGB (P < 0.001) was confined to the 0–0.5-m depth increment, and inter-species differences (P = 0.05) in the mean maximum rooting depth were minimal. For most species, this had increased from ~ 0.1 m in year 1 to ~ 0.3 m by year 5 (Additional file 4). For any individual seedling, the tap-rooted Agathis australis developed the deepest (0.7 m) root, while the plate-rooted systems of Prumnopitys taxifolia, Prumnopitys ferruginea and Alectryon excelsus had the shallowest root systems (~ 0.15 m). By year 5, the mean maximum root depth of each of the trialled species was similar, whether conifers or broadleaved species and tap-rooted or plate-rooted species (Additional file 4).

Similarly, for the first 3 years, 100% of the total BGB (P < 0.001) of all species was confined to within a 0.5-m radius of the root bole. By year 4, only Dacrycarpus dacrydioides, Alectryon excelsus and Vitex lucens had extended their roots to a radius of between 1.0 and 1.5 m from the root bole (Additional file 7). The root mass and percentage of total root mass of Dacrycarpus dacrydioides (0.8 g, 0.4%) and Alectryon excelsus (0.9 g, 0.2%) at this distance were not significantly different from each other but were significantly different from Vitex lucens (3.4 g, 1.7%). Vitex lucens was the only species by year 5 to have extended its root network to a radius of between 1.5 and 2.0 m from the root bole; however, the roots comprised only 8.3 g, 0.8% of its total root mass. Between years 1 and 5, Agathis australis was the only species to have 100% of its root mass confined to within 0.5 m of the root bole. At year 5, > 90% of the total root mass of Dacrycarpus dacrydioides, Dacrydium cupressinum, Prumnopitys taxifolia, Prumnopitys ferruginea and Alectryon excelsus occurred within a 0.5-m radius of the root bole, Podocarpus totara 76.8% and Vitex lucens 65.9% (Fig. 7, Additional file 6).

Root biomass allocation by root diameter size class

Exponential regression analysis of RCD and total root biomass for roots > 1- and > 2-mm diameter size class showed very similar regressions (Fig. 8), with RCD explaining between 73 and 92% of the variation in root biomass for roots > 1 mm and between 78 and 94% for roots > 2 mm (Table 3).

In year 5, fibrous roots accounted for the highest proportion of the total root mass of each of the plate-rooted species Alectryon excelsus (23.5%), Prumnopitys ferruginea (56.4%) and Prumnopitys taxifolia (49.7%) (Fig. 9). Similarly, for the tap-rooted species, fibrous roots accounted for between 28.1% (Dacrycarpus dacrydioides) and 49.3% (Dacrydium cupressinum) of their total root mass. For all species, the percentage of fibrous root biomass decreased as root sizes increased with increasing plant age. The largest root diameter size class (50–100 mm) was recorded for Vitex lucens only and comprised < 2% of its total root biomass. For the remainder of the species, the largest root diameter size class was 20–50 mm and comprised just 0.1% of Prumnopitys ferruginea and 6.8% of the total root mass (largely tap root) of Agathis australis.

BGB/AGB ratio

In year 1, although the BGB/AGB ratio was highest for the tap-rooted Vitex lucens (0.89) and lowest for the plate-rooted Prumnopitys taxifolia (0.25), there was no consistent relationship between this ratio and root type from year to year (Additional file 3). By year 5, Podocarpus totara (0.21) and Prumnopitys taxifolia (0.23) had the lowest BGB/AGB ratio, and although for Vitex lucens the BGB/AGB ratio remained the highest of the trialled species (Fig. 5), it had halved to 0.42 by year 5. Similarly, a decline in BGB allocation with increasing age was apparent in Podocarpus totara (0.47 in year 1, decreasing to 0.21 in year 5) and Prumnopitys ferruginea (0.5 in year 1, decreasing to 0.35 in year 5). Averaged across all eight species, the BGB and AGB was 0.32 and 0.68, respectively.

Root length and root system dimensions (diameter)

Exponential growth analyses of RCD and total root length fitted least well for Podocarpus totara with RCD explaining 46 and 66% of the variation in root length for roots > 1 and > 2 mm, respectively. In contrast, this relationship fitted best for Prumnopitys ferruginea with RCD explaining between 83 and 85% of the variation in root length for roots > 1 mm and > 2 mm, respectively. (Table 3, Fig. 10).

At year 1, none of the species had produced roots > 1-mm diameter. By year 2, Alectryon excelsus and Vitex lucens each produced > 2.5 m of root, significantly more (P = 0.05) than Dacrycarpus dacrydioides (~ 2 m), which in turn was significantly more than that of each of the remaining species at < 1 m of root length (Additional file 4). By year 5, the total root length for each species increased substantially and differences between species became more pronounced. For example, Agathis australis produced the least mean total root length (9.01 m) while Vitex lucens (347.45 m) at the same age produced two orders of magnitude more root length (Additional file 4).

In years 2 and 3 of the trial, and for each of the species trialled, 100% of the total root length occurred within a 0.5-m radius of the root bole, and for Agathis australis, this was the case through to year 5 (Fig. 11, Additional file 7). By year 4, only Dacrycarpus dacrydioides, Alectryon excelsus and Vitex lucens had extended their roots into the 0.5–1.0-m radial sector where Vitex lucens had a significantly greater proportion (P = 0.05) of its total root length (34%) at this radial distance than did Dacrycarpus dacrydioides (19%) and Alectryon excelsus (17%). Similarly, at 1.0–1.5-m radial distance from the root bole, Vitex lucens had significantly more (P = 0.05) of its total root length (~ 5%) than did Dacrycarpus dacrydioides (1%) and Alectryon excelsus (0.4%).

For all species, in years 1 and 2, there was little inter-species difference in mean root system dimensions (i.e. mean maximum root spread). However, by year 5, of the broadleaf species, Vitex lucens had developed a significantly greater (P = 0.05) mean maximum root spread (2.5 m) than had Alectryon excelsus (1.6 m), and its root spread was also significantly greater than that of the largest of the conifer root systems, i.e. that of Podocarpus totara (2.2 m) (Additional file 4). The mean maximum root spread of both Alectryon excelsus and Dacrycarpus dacrydioides (1.7 m), while not different from each other, was significantly larger (P = 0.05) than the more compact root systems of Prumnopitys taxifolia (1.2 m), Prumnopitys ferruginea (1.0 m) and Dacrydium cupressinum (1.1 m), which in turn were not significantly different from each other. The mean maximum root spread of Agathis australis (0.5 m) was significantly smaller than that of the remainder of species trialled (Additional file 4).

Root length by root diameter size class

For all the trialled species, the highest proportion of the total root length of 5-year-old trees was in the 1–2-mm-diameter size class and with the exception of Agathis australis comprised > 70% of their total root length. For Agathis australis, 39% of its total root length comprised roots in the 2–5-mm size class, 20% greater than for the remainder of species (Fig. 12). The percentage of total root length decreased with increasing root size class, and for the 20–50-mm size class, where present, roots comprised ≤ 1.5% of the total root length. None of the trialled species had roots in the > 50-mm diameter size class.

Plant biomass and carbon accumulation

Root system types

Two types of root systems (tap-rooted and plate-rooted) were observed based on the descriptive classification of Hinds and Reid (1957) we recognised.

Of the trialled species that developed a tap root (Agathis australis, Dacrycarpus dacrydioides, Vitex lucens, Podocarpus totara and Dacrydium cupressinum), the seedling radicle of Agathis australis (Fig. 2a) was prominent earliest (3-year-old plants) and extended deeper (0.3 m) than the remainder of the trialled species. By year 5, Podocarpus totara produced the longest laterals extending a distance of 2.2 m, significantly longer (P = 0.05) than all the other trialled species, and of the broadleaved species, Vitex lucens attained the greatest mean root spread (2.5 m) (Additional file 4).

Plate-rooted species include the conifers Prumnopitys taxifolia and Prumnopitys ferruginea and the broadleaved Alectryon excelsus. Their juvenile root systems are compact and consist largely of abundant fibrous and numerous short, gnarly, lateral roots of small diameter size (< 1 mm diameter) somewhat resembling that of Agathis australis but lacking a tap root (Fig. 2b). Of the plate-rooted species, the broadleaved Alectryon excelsus developed the longest lateral roots by year 5 with a spread of 1.6 m, significantly longer (P = 0.05) than the other plate-rooted species (Additional file 4).

Plant allometry

The physiological age of all the species was similar at the time of planting, but there were initial and significant differences in the above-ground metrics among species. Once established, interspecies differences in year-on-year growth remained but became increasingly variable. Payton et al. (2009)⁵ suggested that AGB should be estimated using height and percentage cover for juvenile trees with a DBH < 50 mm. More recently, Mason et al. (2014) used tree height and basal area to determine AGB in New Zealand shrublands while Beets et al. (2012) used a two-factor exponential model which uses DBH or DBH² × height to estimate AGB biomass in mature native forests. Magalhães (2015) suggested that DBH is preferential to RCD as RCD can be affected by root buttress. However, RCD was used in the current study as DBH height (1.4 m) of the juvenile species trialled was not attained until year 4. Indeed, as suggested by Wagner and Ter-Mikaelian (1999), RCD may serve as a better predictor of both the above- and below-ground biomass as it is measured at the intersection of the base of the stem with the root bole. In fact, RCD provided the best fit for tree height and total AGB, and all regressions were highly significant in each of the 5 years of the current trial and for all the trialled species.

To date, generic rather than species-specific allometric equations have been used to estimate the AGB of mature indigenous forest and shrub biomass. Extrapolating these equations to juvenile plants with smaller diameters far outside the range of data used to develop them is questionable however (Claesson et al. 2001; Zianis and Mencuccini 2003). The allometric biomass equations presented in this paper indicate that there are clear differences in growth performance among species, at least at their juvenile stage, and that different fitted coefficients are required. Furthermore, there is a dearth of allometric equations for estimating the below-ground biomass of indigenous species. The absence of empirical data means that a figure of 20% is widely accepted, both locally (Beets et al. 2007) and internationally (Coomes et al. 2002) despite the precision of these root estimates tending to be low (Hall et al. 2001).

Studies that include measurements of both the above- and below-ground growth attributes of the juvenile form of different species have also found RCD to be a good fit for estimating height, AGB and also BGB. For example, Marden et al. (2018; in press) found RCD to be a good fit for estimating all three attributes of 12 of New Zealand’s indigenous riparian species with each species requiring a specific fitted coefficient. Furthermore, Marden et al. (2016) found that two-parameter exponential regressions using RCD was successful in predicting the BGB of five seed lots of Pinus radiata between 1 and 4 years old with RCD explaining between 93 and 98% of the variability in BGB. Similarly, Phillips et al. (2015) found it possible to develop simple relationships between RCD and growth attributes with reasonable r² values for below-ground biomass and total root length.

However, year-on-year growth performance data for the juvenile stage of most of New Zealand’s indigenous species is particularly rare, thus limiting comparisons with other published works. Nonetheless, there is evidence to show that the proportion allocated to BGB decreases with increasing age despite significant year-on-year gains in total plant biomass in many species, both indigenous and exotic (Marden et al. 2005). This concurs with other local (Watson et al. 1995, 1999) and international (Abernethy and Rutherfurd 2001; Easson and Yarbrough 2002) studies where a decrease in the root/shoot ratio for small saplings has been related to an increase in plant size (Kira and Shidei 1967; Walters and Reich 1996) and to an increase in plant height (Cao and Ohkubo 1998). Conversely, Marden et al. (2016) found no clear indication that the root/shoot ratio of 1- to 4-year-old Pinus radiata decreased over this period but instead remained relatively constant. Biomass allocation to components of woody plants is also affected by factors including plant architecture and morphology and climatic and edaphic factors (Ketterings et al. 2001).

In this study, the range of biomass apportioned to below-ground components by year 5 was 23–41%, not dissimilar to that of 12 early colonising indigenous species at the same age with a below-ground biomass of between 24 and 44% (Marden et al. 2005) or for 3-year-old Eucalyptus fastigata (33%), Sequoia sempervirons (29%), Acacia melanoxylon (37%) and Pinus radiata (36%) trialled at the same site (Phillips et al. 2015). Although the below-ground biomass of each of five different 5-year-old clones of Pinus radiata (~ 22%) established on steep hill country (Marden et al. 2016) is at the lower end of this range, it is nonetheless consistent with that established for similar-aged pines planted at a range of sites in New Zealand (Watson and Tombleson 2002, 2004) and likely reflects harsher site and edaphic and climatic factors.

There was no consistent year-to-year difference in the root/shoot ratio between the tap- and plate-rooted species or between conifers and broadleaved species (Additional file 3).

Root metrics and architecture

Published accounts on the architecture of root systems of New Zealand’s commonest and longest-lived native conifer and broadleaved forest species are scarce, descriptive and generally based on observations of partially exposed roots undermined along river banks (Cameron 1963) or of root boles of windthrown trees (Bergin and Steward 2004). At the time of establishing this trial, there were no root-related data (photographic or descriptive) for juvenile specimens of the trialled species. While some of the general findings on root distribution (depth and spread) reported here were anticipated from the results of a previous trial at the same site (Marden et al. 2005), other findings were not. For example, it was anticipated that a high proportion of the total BGB (P < 0.001) would likely remain confined to the 0–0.5-m depth increment, to within a 0.5-m radius of the root bole even in the absence of barriers to root penetration. Also expected was that root biomass and root length would decrease with increasing depth and horizontal distance from the root bole. Similarly, fibrous roots (as a percentage of total root biomass) were expected to decrease as root sizes increased.

Conversely, while excavations of the root systems of juvenile Agathis australis revealed the known presence of a well-developed tap root (Bergin and Steward (2004), the presence of a tap root in Dacrycarpus dacrydioides, Vitex lucens, Podocarpus totara and Dacrydium cupressinum was unexpected and has not to our knowledge been previously reported. Additionally, their many leading lateral roots, though initially of small diameter, were highly symmetric; thus, few areas of soil were left totally devoid of any roots. These species typically had a high number of roots per unit volume of soil and, as expected for juvenile plants, the bulk of their root mass and total root length was confined to a shallow depth and within close proximity to the root bole. By comparison, the root systems of the juvenile plate-rooted species (including the conifers Prumnopitys taxifolia and Prumnopitys ferruginea and the broadleaved Alectryon excelsus) typically had fewer laterals roots but were of larger diameter and with limited branching tended to developed in preferential directions. Thus, the number of roots per unit volume of soil was lower than for the tap-rooted species with significant areas of soil remaining devoid of root mass. This was also observed during the excavation of root systems of non-native forest species (Phillips et al. 2015) and non-native Populus and Salix species at the same trial site (Phillips et al. 2014). These results suggest that interspecies differences in the distribution of roots and their overall root system architecture are likely to be inherent rather than due to seedling propagation, planting practices or site factors given that there were no observable differences in the physical attributes of the soil across this essentially flat terrace surface and competing influences from adjacent well-spaced plants were unlikely.

Conversely, the highly asymmetric root architecture of juvenile Pinus radiata reported by Marden et al. (2016) is considered typical of tree root systems developed on steep slopes where the greatest proportion of the total root length trends in diagonally opposite directions previously referred to as bilateral fan-shaped architecture (Chiatante et al. 2003). Here, the larger and stronger lateral roots develop in the upslope (Schiechtl 1980) under compression (Stokes and Guitard 1997) and provide the greatest contribution to tree stability/anchorage (Sun et al. 2008). Roots located in the downslope direction develop under tension, and are therefore smaller and weaker, and likely contribute less to tree anchorage. Nicoll and Ray (1996) reported that prevailing winds can influence the distribution of the structural roots of Sitka spruce (Picea sitchensis (Bong.) Carr.) where, relative to the prevailing wind, there were more roots located on the leeward side than on the windward side of trees. Marden et al. (2016) considered the development of the bilateral fan-shaped distribution of structural roots of Pinus radiata to have been influenced by the predominance of northerly winds as an adaptation to improve the rigidity of the soil-root plate and to counteract their vulnerability to windthrow.

The openness and exposure of the conifer and broadleaved species trial site to strong winds resulted in stem breakage of Alectryon excelsus, and Vitex lucens was defoliated during periods of heavy frost. However, site factors are unlikely to have contributed to inter-species differences in their overall growth performance as all the trialled species tolerate a wide range of environmental conditions. Furthermore, with weed control, unlimited water availability, a fertile alluvial soil and the absence of barriers to root development, the resultant growth performance could be considered as optimal for these species. However, as is common with nursery-raised juvenile plants, factors such as wrenching, root trimming and planter bag constriction likely stunt the rate of root biomass accumulation, root spread and depth within the first 3 years following planting (Marden et al. 2005). To what extent minor root binding affected the eventual distribution of root diameter size classes and spatial pattern of root development of the species reported in this paper remains uncertain. Nonetheless, it was clear that the timing and proportion of total plant biomass allocated to roots and root bole did not increase significantly until years 4 and 5, and this appears to be related to the period (years after planting) that plants require to establish their root systems sufficiently to sustain an increase in growth rate. With the exception of the broadleaved and plate-rooted Alectryon excelsus, the tap-rooted and predominantly coniferous species allocated more of their total BGB to the root bole than did the remaining plate-rooted species or the tap-rooted Agathis australis. That said, there is also likely to be a relationship between plant size and the rate of root and root bole biomass accumulation with the smallest and slowest growing of the conifers Agathis australis, Prumnopitys taxifolia and Prumnopitys ferruginea amassing the least BGB.

Between years 1 and 3, there was no consistent relationship between the diameter of the canopy and root system dimensions (diameter) of tap- versus plate-rooted species or between conifer and broadleaved species. However, by year 4 and thereafter, the diameter of the root systems of the broadleaved Alectryon excelsus and Vitex lucens and the conifer Dacrycarpus dacrydioides exceeded the width of their respective canopies suggesting that the ratio of root spread to canopy dripline may be different for different species at the same age. From the few available descriptions of root system diameters of mature trees, it is evident that the lateral roots of conifers Dacrydium cupressinum and Podocarpus totara may extend > 9 m from the root bole (Bergin and Gea 2005, 2007). Similarly, the lateral roots of mature Prumnopitys taxifolia are often seen exposed on the ground surface and extending to > 19.5 m from the root bole before disappearing into the soil (Allan 1926), and thus, for the more conical-shaped conifers, their lateral roots likely extend beyond the canopy dripline.

The high concentration of roots (mass and length) of both the tap- and plate-rooted species at shallow depth was also anticipated. General observations of the root systems of mature plate-rooted indigenous species confirm that their root systems are largely confined to the uppermost metre or two of the soil with the prevalence of shallow rooting in forest soils likely reflecting dependence on nutrient cycling through litter fall (Yeates 1924; Cranwell and Moore 1936; Wardle 2002). Furthermore, observations of the root systems of the mature, plate-rooted Dacrycarpus dacrydioides indicate that their lateral structural roots are shallow and often sub-aerial as an adaptation to imperfectly drained soils and/or wetter flatland sites including alluvial floodplains and terraces with a periodic high water table dominated by pasture grasses (Wardle 2002). Alternatively, as suggested by Nicoll and Ray (1996), a shallow and extensive lateral root network may also be an adaptation to improve the rigidity of the soil-root plate and counteract the increasing vulnerability to windthrow as trees grow taller. Similarly, the development of an extensive network of lateral structural roots in Populus and Salix species makes them the preferred species for soil conservation and slope reinforcement on poorly drained sites such as earthflows (Phillips et al. 2014).

The maximum rooting depth recorded for an indigenous forest species is for a mature tap-rooted Agathis australis. Unless impeded by rock or a hard soil pan, tap roots have been recorded at depths in excess of 2 m (Bergin and Steward 2004), and ‘sinker roots’ have been observed to descend from large, lateral roots to a depth of 4 m and terminate in a network of smaller roots (R. LLoyd, pers. comm. cited in; Bergin and Steward 2004). Agathis australis is predominately found growing in the more sub-tropical regions of New Zealand so the early development of a deep tap root and relatively large biomass allocation to roots should be beneficial to surviving droughts as suggested by Becker and Castillo (1990). It has also been suggested that a larger (and early) allocation to root biomass is beneficial for tree establishment (T Poulson, pers. comm. cited in Coutts 1987).

Biomass and carbon accumulation

Nationally, there is 1.45 Mha of marginal pastoral land suitable for reforestation by indigenous trees and shrubs (Trotter et al. 2005). The industry standard for establishment plantings of non-native species (principally Pinus radiata) on marginal land varies considerably depending on the class of land and its erosion severity (National Water and Soil Conservation Organisation 1976). In the highly erodible East Coast region of the North Island, the planting density is generally around 1000–1250 stems per hectare (spha). The establishment of a mixed indigenous forest comprising 125 of each of the eight species trialled here (1000 spha) would potentially amass ~ 1.5 t ha⁻¹ within 5 years of planting, which would generate a forest carbon stock of 2.5 t CO₂ ha⁻¹. With the exception of Agathis australis, the remainder of the conifers (mainly tap-rooted species) individually accumulated significantly more total plant biomass than did any of the plate-rooted species, while the broadleaved Vitex lucens (also tap-rooted) accumulated the largest total biomass of any of the trialled species (Table 4, Additional file 3). Selection of a species mix comprising 200 of each of the best performing of the tap-rooted conifers (Dacrycarpus dacrydioides, Dacrydium cupressinum, Podocarpus totara) and the broadleaved Vitex lucens, together with the best performing of the plate-rooted broadleaved Alectryon excelsa would amass ~ 2.3 t ha⁻¹, a forest carbon stock of 3.8 t CO₂ ha⁻¹. Furthermore, an early colonising indigenous shrubland species mānuka (Leptospermum scoparium (Forster et Forster f.)) established at the same density would amass ~ 3.7 t ha⁻¹ resulting in a forest carbon stock of 6.1 t CO₂ ha⁻¹ over the same 5-year period (Marden and Lambie 2016). Thus, the planting of Leptospermum scoparium on marginal land would store an additional ~ 2.3 t CO₂ ha⁻¹ over an equivalent period of time than that gained from planting an equivalent density of the best performed of the trialled broadleaved and conifer species. Interestingly, the carbon stock for areas of indigenous forest 5 years after establishment (planting), or from the date when land-use change to reversion commences, is estimated at 7.8 t CO₂ ha⁻¹ (Ministry for Primary industries 2017). By implication, to achieve a similar level of carbon stock within this time frame would for areas of new plantings require ~ 2000 spha of the best performed of the trialled broadleaved and conifer species. Similarly, for areas set aside to permit passive reversion to occur would require an equivalent density of mixed broadleaved and conifer species, and for areas actively planted and managed for mānuka honey production, 1200–1300 spha would be required.

For the best performed of the trialled species, a monoculture of broadleaved Vitex lucens planted at 1000 spha could potentially achieve a level of forest carbon stock of 7.07 t CO₂ ha⁻¹ by year 5 (Table 4), which approximated to the 7.8 t CO₂ ha⁻¹ estimated by the Ministry for Primary Industries (2017) for new areas of planted and reverting shrubland. However, irrespective of the species mix and planting regime adopted for new areas of indigenous forest establishment, the potentially achievable forest carbon stock by year 5 is likely to be an order of magnitude less than the ~ 91.6 t CO₂ ha⁻¹ of forest carbon stock recorded for 4-year-old Pinus radiata (from Marden et al. 2016) and the 77 t CO₂ ha⁻¹ estimated by the Ministry for Primary Industries (2017) for 5-year-old post-1989 plantings in the East Coast region.

Notwithstanding, the planting of indigenous conifers and broadleaved species on erodible marginal land will ultimately afford a more continuous and permanent forest sink for carbon over an active growth phase that extends from 150 to 500 years (Hall 2001; Hinds and Reid 1957) that is consistent with the prolonged effort likely required to effect significant reductions in atmospheric CO₂ levels.

Towards implementation

The second objective of this trial was to assess the potential effectiveness of the trialled species for erosion control on newly afforested/reforested marginal pastoral land and/or as replacement species best suited to providing long-term stability to areas of exotic forest deemed at greatest risk to storm-related erosion and under consideration for retirement in the future.

The well-developed literature on the geomechanical effects of vegetation on hill slope stability clearly indicate that tree roots reinforce soils, contribute to soil strength and stabilise eroding hillslopes. The mechanical interactions of woody roots with the soil medium play an important role in tree anchorage and thus in the prevention of slope failure. Thus, one of the most important traits is rooting depth, which, when soil conditions are not limiting, is species dependent. The most effective slope stabilisation results when root development occurs at different depths in the soil profile (Schiechtl and Stern 1994; Czernin and Phillips 2005). This has been observed in exposed root boles of mature Dacrydium cupressinum and Agathis australis.

In New Zealand, shallow landslides on typically steep slopes underlain by Tertiary-aged bedrock are generally ≤ 1 m deep (Marden et al. 1991; Page et al. 1994) so roots must cross the basal shear plane in order to anchor the soil into more stable substrate and stabilise a slope against the initiation of shallow landslides (Wu et al. 1979). This study shows that, for most of the tree species examined, the root systems are concentrated predominantly in the upper soil profile with mean maximum root depths by year 5 ranging between 0.3 and 0.4 m. Furthermore, at this stage of their development, there was no significant difference in this metric among individual conifer species, between conifer and broadleaved species or between plate-rooted and tap-rooted species. The results also indicate that there is a rapid decline in roots with increasing depth and distance from the stem, that there are interspecies differences in root distribution and that each species allocates differing proportions of their total biomass to roots at different stages of growth, a finding similar to other local (Watson et al. 1995, 1999) and international (Abernethy and Rutherfurd 2001; Easson and Yarbrough 2002) studies on plant root distribution. This, together with the few published reports on the root depth of some of New Zealand’s tallest native conifer and hardwood forest species (Cameron 1963), with the exception of Agathis australis, indicate that the rooting depth of mature trees rarely exceeded 2–3 m. By implication, their comparatively shallow rooting depth will likely be a limiting factor in their ability to contribute to deep-soil reinforcement. Similarly, the ability of widely spaced plantings to prevent the occurrence of shallow landslide diminishes as root distribution decreases with increasing distance from the root bole.

Interspecies differences in root distribution (depth and lateral spread) have implications for the planting densities required to provide full, near-surface root occupancy of the soil and/or to maximise the root density to the depth requirement at specific sites (Phillips et al. 2012). For example, a dense lateral network of woody roots predominantly confined to the upper soil horizons often forms a membrane that stabilises shallow soils (Schmidt et al. 2001) and significantly reduces shallow landslide potential on steep slopes (Ekanayake et al. 1997), while the larger tree roots can provide reinforcement across lateral planes of weakness bounding potential slope failures (Sidle et al. 1985). However, deeper soil mantles (> 5 m) benefit little from such reinforcement as root density decreases dramatically with depth and few large roots are able to anchor across the basal shear plane (Stokes et al. 2008). The only benefit of root strength to the stabilisation of deep-seated landslides, e.g. earthflows and slumps, is when very large lateral woody roots cross planes of weakness along the flanks of potential failures (Sidle and Ochiai 2006). However, such reinforcement would only benefit deep-seated landslides of small area extent whereas lateral roots also offer protection against most shallow landslides (Swanston and Swanson 1976).

Larger diameter roots at greater soil depths would be desirable to stabilise the deeper forms of mass movement; thus, root diameter is another important consideration on landslide-prone slopes as thick roots reinforce soil in the same way that concrete is reinforced with steel rods, and the thicker the root, the greater its ability to penetrate soil (Clark et al. 2008). The most common barriers limiting the development of a deep rooting system include a permanent high water table, bedrock interface or a cemented iron pan and stoniness and depth of colluvium. In general terms, the root:shoot ratios of seedlings are depressed on such sites. As previously found for other New Zealand woody species, root depth was not necessarily correlated with tree age but rather to the stoniness and depth of slope colluvium (Watson et al. 1995).

In terms of root anchorage, tap-rooted species will, in time, likely provide a higher level of reinforcement directly under a tree and to a greater depth than for plate-rooted species. Thus, for tap-rooted species, root reinforcement quickly tapers off laterally as root density declines with increasing distance from the tree. In contrast, plate-rooted species with their widespread development of a lateral root network will likely provide a high level of near-surface soil-root reinforcement but nonetheless remain more susceptible to uprooting than will tap-rooted species because as the number of roots decline with increased depth (Gray and Sotir 1996) so does root anchorage rapidly decline (Dupuy et al. 2005; Norris et al. 2008).

As a consequence of their shallow-rooted habit, many of New Zealand’s conifer and broadleaved trees will have limited effectiveness in floodplain reaches where channel hydraulic conditions are likely to result in the undercutting of stream banks to a depth greater than that of the maximum penetration depth of root systems (Marden et al. 2005). If the potential for streambed degradation exists, additional protection in the form of structural materials (e.g. gabion baskets and rip rap) will be required along the toe of banks and to some depth below normal streambed level. The limitations of root depth aside, some of the trialled conifer and broadleaved species have adaptations suitable for floodplain restoration. Alluvial communities of some of New Zealand’s oldest podocarps (Podocarpus totara, Dacrycarpus dacrydioides) are tolerant of wet soils and swampy sites subjected to periodic flooding and seasonally high water tables. They have the ability to produce a new root system after flooding and inundation with river silt, and Podocarpus totara is able to produce tiers of adventitious roots from the buried part of the stem should the tree topple or break. Dacrycarpus dacrydioides, however, is intolerant of siltation if greater than ~ 60 cm (Foweraker 1929; McSweeney 1982; Campbell 1984). Many species, including Dacrydium cupressinum, Podocarpus totara and Agathis australis, are equally tolerant of harsh growing conditions on steep slopes with skeletal soils deficient in nutrients that are prone to drying during seasonal droughts.

Although native conifers and broadleaved trees have not traditionally been used specifically for soil conservation, their successional role meets most of the requirements for use in revegetation where long-term erosion control is required (Pollock 1986). It is also important to consider that, in addition to its erosion-control function, a species mix may have greater associated ecological, cultural, aesthetic and economic benefits than a single species stand. In addition, most are sufficiently diverse in rooting form to stabilise slopes prone to shallow landslides and dry ravelling, while those with the limitations of root depth will be less effective in stabilising the deeper forms of mass movements such as earthflows and slumps as their failure plane tends to be deeper than the vertical limit of root penetration. Plant density and species mix are also key factors in determining plant effectiveness; thus, options that promote the quickest canopy closure and root development at all levels of the soil profile are likely to be the most effective in promoting slope stability (Phillips et al. 2001). The faster the soil is occupied by roots, the greater the reinforcement of the soil and the greater the chance of limiting landslide initiation on erosion-prone slopes and dry ravelling of exposed soil and rock. As an example, soil occupancy by the larger root systems of the broadleaved Vitex lucens and Alectryon excelsus would occur sooner (~ 5 years after planting) than the best performed of the conifers (Podocarpus totara) tested under optimal growth conditions and using the mean maximum root spread at year 5 as an indicator of the rate at which soil-root reinforcement might be considered as being effective against erosion. Conversely, the more compact root systems of Prumnopitys taxifolia, Prumnopitys ferruginea and Dacrydium cupressinum would take longer (~ 10–15 years) to be effective and the significantly smaller root system of Agathis australis, the longest. Phillips et al. (2015) compared the lateral root spread of exotic species including alder (Alnus rubra), cherry (Prunus serrulatus), blackwood (Acacia melanoxylon) and redwood (Sequoia sempervirons) with reported values for indigenous riparian species (Marden et al. 2005), for mānuka (Marden and Lambie 2016) and for Pinus radiata at a similar age (Watson and Tombleson 2002, 2004; Marden et al. 2016). The conclusion drawn was that the species with the larger root system dimensions would provide earlier soil-root reinforcement and thus be more effective in mitigating the initiation of shallow landslides. Similarly, using root depth as an indicator of species potential for stabilising deeper-seated forms of erosion (e.g. rotational slumps and larger earthflows), tap-rooted species (e.g. Agathis australis) have the potential to develop thicker and stronger roots able to penetrate across the basal shear plane to provide a deeper level of reinforcement, albeit only for the smaller deep-seated failures, than would plate-rooted species. Nonetheless, as was evident before the mature indigenous forest cover was removed from hill-country areas, given time to mature, these same species will slow down but may not completely arrest many of the deeper-seated forms of mass failure.

For many restoration sites, the strategy adopted should ideally include a mix of early colonising and successional species with plate- and tap-rooted characteristics. On less exposed sites, a mixture of shrub and tree species can be planted concurrently with the longer-lived conifer and broadleaved species interplanted at near final spacing (Bergin and Gea 2005, 2007). Another option that has found favour with many community groups is to restrict the number of colonising species (three or four) and then carry out follow-up planting with the late succession, longer-lived species after a few years to fill in any gaps created by failures from the initial planting (Fred Lichtwark pers. comm.).

During selection of suitable plant materials, the capacity of species to match specific site conditions, e.g. those that are long lived and can overcome potential limitations to successful establishment such as overcrowding resulting in suppressed growth (less is often best in the long term), frost, fungal and insect attack, browsing, wind throw and stem breakage, must also be taken into account. Potential limitations to a successful establishment outcome of the trialled species, particularly during the early growth period, include susceptibility to frost (e.g. Alectryon excelsus, Vitex lucens, Dacrydium cupressinum, Agathis australis), insect damage (e.g. Podocarpus totara, Dacrydium cupressinum, Dacrycarpus dacrydioides, Agathis australis), fungi (e.g. Agathis australis, Podocarpus totara), browsing feral mammals (e.g. Dacrydium cupressinum, Podocarpus totara, Agathis australis) and stem snap (e.g. Alectryon excelsus) (Bergin and Gea 2005, 2007).