Introduction

Wood density is an ecologically and economically important trait in plants. High wood density increases resistance to physical damage (Niklas 1992) and pathogen resistance (Hillis 1987; Augspurger and Kelly 1984) and thereby contributes to plant survival. Additionally, wood density is an important parameter for estimating the biomass of trees and is used to calculate the carbon stocks in forests (Fearnside 1997). Within a species wood density can vary across environmental gradients, as its variation depends on the anatomy of the wood, which is in turn linked to functions such as water transport, mechanical strength and growth (Chave et al. 2009).

The density of wood in many species is largely determined by fibres, which are thick-walled cells composed of cellulose and lignin, that provide mechanical support (Barnett and Bonham 2004; Beck 2010). Other tissue types can also influence wood density. The abundance of xylem vessels, which have lumen areas for water transport, decreases wood density. Previous studies over a range of species have shown increasing wood density correlates with decreasing xylem vessel lumen area (Hacke et al. 2001; Preston et al. 2006).

Wood density has also been linked to tree growth rates. In a survey of 21 rain forest species from Malaysia King et al. (2006) found tree growth rates decreased with increasing wood density. The relationship between wood density and tree growth rates reflects the presence of large xylem vessels, which allow for high rates of transpiration and photosynthetic carbon gain that supports large canopies and high growth rates (Enquist et al. 1999; King et al. 2006).

Mangrove forests dominate the coasts of tropical and subtropical coastlines, where they are exposed to a range of physical disturbances, including ocean waves, strong winds and tidal currents. Mangroves provide valuable coastal protection services (Mazda et al. 2006) and are increasingly recognized for their role in carbon sequestration (Mcleod et al. 2011). The density of mangrove wood is key to the provision of these ecological services, yet the anatomical basis of variation in wood density in mangroves is poorly known. In this study we assessed variation in wood density of the mangrove species, Avicennia marina.

The Indo-pacific mangrove species A. marina has the widest geographical distribution of all mangroves with a latitudinal distribution from ~25°N to 38°S. Avicennia marina reaches its southernmost distribution at Corner Inlet, Victoria, Australia and in the North Island of New Zealand (Duke 2006). The species tolerates a wide range of soil water salinity and is the dominant species where aridity and/or cool temperatures limit the productivity of mangroves (Morrisey et al. 2010). The wide distribution of A. marina has been linked to its special wood structure, characterized by successive cambia (Tomlinson 1995; Robert et al. 2011). This peculiar wood structure comprises consecutive bands of xylem, interspersed with phloem strands that are connected by a band of parenchyma tissue (Schmitz et al. 2007, 2008). Based on the occurrence of this wood structure in tree species that occur in arid or saline habitats (i.e. only 34 families are known to have this wood structure), it has been proposed that successive cambia are associated with tolerance to water deficits (Robert et al. 2011). The presence of phloem in the successive cambia may have implications for wood density.

Here we assessed variation in wood density of A. marina in two locations with different climatic conditions and across an intertidal gradient. Our goals were to (1) assess the variation in wood density between and within locations, (2) investigate the contribution of anatomical features to variation in wood density and (3) test for linkages among variation in wood density, anatomy and tree growth rates in A. marina.

Methods

Site descriptions and sample collection

Our study compares the density and structure of the wood of A. marina and its correlation with tree growth rates at two sampling locations where A. marina is dominant, the Firth of Thames, New Zealand (NZ) and Giralia Bay, Exmouth Gulf in Western Australia (WA).

The Firth of Thames (37.18°S, 175.4°E) is an embayment of about 19-km width in the northern part of the Hauraki Depression, NZ (Hochstein and Nixon 1979). In this site, an extensive mangrove forest with varying forest structure (ranging between ~0.3 to 3 m in height) has developed since the 1950s (Lovelock et al. 2010). The climate is wet, with a mean annual rainfall of 1,400 mm year−1 and cool, with a mean annual air temperature of 15 °C (National Institute of Water and Atmospheric Research (NIWA) 2011). Our second sampling location at Giralia Bay in the Exmouth Gulf, WA (22.4°S, 114.3°E) is arid, with a mean annual rainfall of 300 mm year−1 and warm, with a mean annual air temperature of 25 °C. Cyclones occur every two to three years in the Exmouth region causing short periods of intense rainfall and flooding (Lovelock et al. 2011) and wind gusts in excess of 90 km/h (Australian Bureau of Meteorology 2011).

In January 2007, we collected five mature stems from each of three sites across the intertidal zone in NZ. The sites encompassed trees of differing sizes and ages (Lovelock et al. 2011). The forest at the Firth of Thames, NZ is older towards the landward side (trees recruited in ~1978) and younger on the seaward edge (trees recruited in ~ 1996). In the NZ forests tree size varies in a typical manner for mangroves (Feller et al. 2010), with tall trees on the seaward, more frequently inundated site (NZ-Seaward) to scrub forests ~0.3 m in height in the landward, less frequently inundated site (NZ-Landward-1, 350 m from the seaward edge). The older 1978 forest is a relict fringing forest (NZ-Landward-2, 620 m from the seaward edge) that has tall trees, but is inundated less frequently due to sediment accretion and forest development in a seaward direction over time. At Giralia Bay in WA, we collected wood from six mature stems from the seaward edge of the forest (WA-Seaward) in October 2008. Similar to the fringing site in NZ, the seaward fringing forest at Giralia Bay is inundated by the tide twice daily.

At each site (three at NZ and one in WA), we assessed the porewater salinity. We collected water samples from 30 cm depth using a suction device (McKee et al. 1988) and analysed them with a handheld refractometer (W/ATC 300011, SPER Scientific Scottsdale, USA).

Measurement of wood characteristics

All our measurements were made in the 3–4 outermost layers of wood (Fig. 1a). Wood density was measured for each tree from 1 cm × 1 cm × 2 cm rectangular prisms cut from the outer layers of the wood. Wood density was calculated as the ratio of the oven-dry mass of the wood sample divided by the mass of water displaced by its fresh (green) volume (Chave et al. 2006). Using oven-dried (60 °C) and sanded wood discs (Fig. 1a, b), the number of growth layers, defined as ontogenetic units of phloem and xylem, was counted using an Olympus SZH10 microscope (Tokyo, Japan). A growth layer was counted if it crossed a line drawn from pith to bark along the maximum radius of the stem disc (Schmitz et al. 2007). The proportion of phloem per growth layer was calculated as the width of the phloem band divided by the total width of the growth layer, including both xylem and phloem, measured along the radius of the stem disc.

Fig. 1
figure 1

a Sanded wood disc of Avicennia marina. Inset shows the last layers in which we sampled the wood. b We defined a “layer” as a complete band of xylem (X) and phloem (P) that is composed of conjunctive tissue including phloem strands. c Transverse wood section of Avicennia marina. Insets show how we measured double fibre wall thickness and xylem vessel axes to calculate fibre wall thickness and xylem vessel diameter (Lewis 1992)

Full size image

Given it can be difficult to resolve small vessels from sanded wood discs, ~1 cm3 cubes were cut from the discs to make transverse microsections (Fig. 1c) that were double stained with Safranin–Alcian blue. We photographed the sections with an Olympus BX61 microscope (Tokyo, Japan). From the photographs we measured 100–200 xylem vessels and 100–200 fibres for each stem. Xylem and phloem area was measured per layer. The number of vessels was counted and the perpendicular axes of the vessel lumina were measured (Fig. 1c). All the measurements were done using Image Pro Plus version 5.0.1 (MediaCybernetics, Maryland, USA).

We calculated the xylem vessel diameter (D) following the ellipse equation of Lewis (1992) (Eq. 1), and the total vessel lumen area (XA) accounting for vessel density (VD) and vessel area (VA) (Eqs. 2, 3, 4).

$$ D = \sqrt {\frac{{2a^{2} b^{2} }}{{a^{2} b^{2} }}} $$
(1)
$$ XA = \overline{VAVD} $$
(2)

with vessel density,

$$ {\text{VD}} = \frac{{\# {\text{Vessels}}}}{{{\text{mm}}^{ 2} }} $$
(3)

and vessel area,

$$ {\text{VA}} = \pi ab0.25 $$
(4)

Fibre wall thickness was calculated as the radial distance from fibre lumen 1 to fibre lumen 2 divided by two (Fig. 1c).

Tree growth rates

For the A. marina trees from NZ the pith of each sampled tree was dated using bomb-pulse radio carbon dating with Accelerator Mass Spectrometry (AMS) 14C analysis (Table 1) (Lovelock et al. 2010). Wood samples were dated with a resolution from 1 to 2 years, which is possible given the significant changes in atmospheric 14C after 1955 (when the last bomb-pulse was released) (Hua and Barbetti 2004; Hua 2009). Growth rates were estimated as stem radius divided by tree age. For the trees from the WA-Seaward site, growth rates were estimated from the mean circumference increment in the period April 2008–August 2010, as recorded with 27 mechanical dendrometer bands and converted to radius increment per year.

Table 1 Description of the characteristics of A. marina trees from the seaward and landward sites of the mangrove forests in New Zealand and Western Australia
Full size table

Data analyses

Differences in wood density, xylem vessel diameter and the proportion of phloem per growth layer between sites were tested using a Kruskal–Wallis test and a Dunn’s post hoc test. These analyses were performed with the software package Prism 5.0 (La Jolla, CA, USA).

To analyse the influence of anatomical characteristics on variation in wood density we used linear regression and stepwise multiple regression analyses. Linear regression analyses were performed using the package lmodel2 (Major Axis) from R (The R Foundation for Statistical Computing). Bootstrap analyses were performed to assess the suitability of the sample size. Confidence intervals of the slope and the intercept were calculated for the linear regressions using lmodel2 and the bias was calculated for the multiple regressions with the R software (bias after bootstrap in all cases was <0.001). These tests indicated that our sample size was adequate and the results were independent of sample size. Prior to analysis we tested for normality using the D’Agostino-Pearson normality test. Independent variables used in the stepwise multiple regression were not correlated.

Results

Wood density was higher in WA than in NZ (Fig. 2; n = 19, p < 0.005). Between the NZ sites wood density was similar in the tall trees from NZ-Seaward and NZ-Landward-2, but wood density was significantly lower in the NZ-Landward-1 scrub trees (Fig. 2).

Fig. 2
figure 2

Mean wood density of trees at sampling sites in New Zealand and Western Australia. Values are the means and standard deviations from 3 to 6 individual trees. Different letters indicate that the means were significantly different (p < 0.05, n = 19)

Full size image

Our analysis of the anatomical factors that contribute to wood density showed that over all our sites wood density was positively related to mean xylem vessel diameter and total vessel lumen area (Table 2; Fig. 3a, b). Variation in total vessel lumen area can be due to variation in vessel density and/or variation in vessel size. Our examination of xylem vessel density found that vessel density varied from 57 ± 13 vessels mm−2 (mean ± standard deviation) in NZ-Seaward to 64 ± 13 vessels mm−2 in WA-Seaward but these values were not significantly different between sites. Therefore, variation in total vessel lumen area between locations was due to variation in xylem vessel diameter.

Table 2 Parameters of the regression analysis of wood density, xylem vessel diameter and growth rates
Full size table
Fig. 3
figure 3

The relationship between a wood density and xylem vessel diameter, b wood density and total vessel lumen area, c wood density and fibre wall thickness and d wood density and the proportion of phloem per wood layer expressed as a percentage (phloem%/layer) (Details of regression lines are presented in Table 2). Different symbols represent different sites and different positions in the intertidal: WA-Seaward (closed circles), NZ-Seaward (open squares), NZ-Landward-1 (open diamonds) and NZ-Landward-2 (open triangles)

Full size image

We also observed that A. marina trees with high wood density had thicker fibre walls (Table 2; Fig. 3c) and a lower proportion of phloem per growth layer (Table 2; Fig. 3d). The proportion of phloem per growth layer was highest in landward forests from NZ (Fig. 4). The three components together, total vessel lumen area, fibre wall thickness and the proportion of phloem per growth layer, explained 76 % of the variation in the wood density of A. marina (Table 3).

Fig. 4
figure 4

The proportion of phloem per growth layer expressed as a percentage (Phloem%/Growth Layer) over the sampling sites in New Zealand and Western Australia. Values are the means and standard deviations from 5 to 6 individual trees. Different letters indicate that the means were significantly different (p < 0.05, n = 21)

Full size image
Table 3 Results of the stepwise multiple regression analyses for wood density
Full size table

Tree growth rates increased with larger xylem vessels and were associated with higher wood density over all sites (Table 2; Fig. 5a, b). However, the regression of growth rates and wood density was significant only for the NZ sites. In contrast, over all sites wood density and xylem vessel diameters were significantly positively correlated with tree size (Fig. 6).

Fig. 5
figure 5

The relationship between a tree growth rate and xylem vessel diameter, and b tree growth rate and wood density (Details of regression lines are presented in Table 2). Different symbols represent different sites and different positions in the intertidal: WA-Seaward (closed circles), NZ-Seaward (open squares), NZ-Landward-1 (open diamonds) and NZ-Landward-2 (open triangles)

Full size image
Fig. 6
figure 6

The relationship between a wood density and stem radius and b xylem vessel diameter and stem radius for the mangrove Avicennia marina (Details of regression lines are presented in Table 2). Different symbols represent different sites and different positions in the intertidal: WA-Seaward (closed circles), NZ-Seaward (open squares), NZ-Landward-1 (open diamonds) and NZ-Landward-2 (open triangles)

Full size image

Discussion

Over our contrasting sites we found that wood density of A. marina varied (Fig. 2). In the same habitat (seaward fringe) A. marina tended to have higher mean wood density in WA than in NZ. In different sites within the same location in NZ, landward scrub mangroves (NZ-Landward-1) had lower wood density than tall trees in seaward fringing forest (NZ-Seaward) and those in a taller landward forest (NZ-Landward-2). Using the variation in wood density over all sites we found that wood density increased with total vessel lumen area and fibre wall thickness but was negatively correlated with the proportion of phloem per growth layer of wood.

The positive relationship between wood density and total vessel lumen area and xylem vessel diameter observed in A. marina is in contrast with previous studies that have found high wood density is associated with low total vessel lumen areas and small vessel sizes (Preston et al. 2006; Jacobsen et al. 2007). However, by increasing fibre wall thickness, A. marina maintains high wood density even with large xylem diameters (Fig. 3a, c). By investing in thick fibre walls and reducing the proportion of phloem per growth layer in wood, A. marina could maintain both high water conductivity and mechanical strength. Other studies have also observed that high wood density is not always associated with low xylem lumen area, indicating that trees may achieve high wood density using a range of anatomical strategies (Jacobsen et al. 2005; Martínez-Cabrera et al. 2009).

Wood density of A. marina increased with increasing fibre wall thickness as has been observed in other tree species (Jacobsen et al. 2005; Martínez-Cabrera et al. 2009). The walls of fibres of A. marina are very thick compared with other trees species and their fibre lumina is almost completely lost (International Association of Wood Anatomists (IAWA) 2012). For example, in this study mean fibre wall thickness in WA-Seaward trees was 5.27 ± 0.55 μm (mean ± standard deviation) in A. marina compared with 2 μm in Rollinia rensoniana (Barajas-Morales 1985) The thick-walled fibres of A. marina may increase resistance to breakage by wind and tidal waters during storms, even in rapidly growing trees with large total vessel lumen areas. In WA the large vessels observed in A. marina trees occurred within a dense fibre matrix. In addition to protection from breakage and attack from wood boring invertebrates, fibres add strength to xylem walls that may increase their resistance to collapse under the high tensions that may occur on the xylem water column with water deficits (Jacobsen et al. 2005).

We observed tree growth rates increased with increasing xylem vessel diameter and wood density at the NZ site (Fig. 5). Stronger correlations (over all sites) were observed between xylem vessel size and stem diameter and wood density and stem diameter (Fig. 6), possibly because stem diameter is an integrated measure of tree growth over the entire life time of trees compared with growth rates calculated from changes in stem diameter over 2.3 years of observations, as they were for the WA trees. The relationship between large lumen areas and fast growth rates and tree size is predicted by theoretical analyses of tree growth (Enquist et al. 1999) and has been observed in other species (Poorter et al. 2010; Russo et al. 2010).

Low wood density of A. marina in our study was strongly associated with high proportions of phloem tissue per growth layer in the wood (Fig. 3d). Phloem is a major constituent of A. marina’s wood tissue and results from secondary growth via successive cambia (Fig. 1b, c). Successive cambia have been described in shrubs and tree species of which ~84 % have been found to grow under conditions where drought is common (Robert et al. 2011). Phloem within the wood may be a site for starch depolymerisation, which is the basis of the mechanism for refilling of embolized xylem vessels (Nardini et al. 2010, 2011) and therefore phloem in wood has been proposed to be important in tolerating salinity stress in A. marina (Robert et al. 2011) and drought stress in other species (Salleo et al. 2004). The proportion of phloem per growth layer in the wood varied across the intertidal gradient in NZ, being significantly higher in landward trees compared to seaward trees (Fig. 4). Additionally, previous studies found higher proportions of phloem per growth layer in landward A. marina trees from Kenya (Schmitz et al. 2007, 2008). Our results indicate a high level of phloem in wood is associated with reduced wood density. This may have negative consequences for withstanding damage from wind and waves and also pathogens and pests. Trees growing in suboptimal habitats (high intertidal) with low wood density may have a greater likelihood of physical damage than trees growing in more seaward sites.

Conclusions

Wood density of A. marina varied between locations and across an intertidal gradient in NZ. High wood density of A. marina was associated with large xylem vessel diameters (and total lumen area), thick-walled fibres, low proportion of phloem in the wood, faster growth rates and larger tree sizes. In contrast, low wood density was associated with large areas of phloem within the wood, smaller xylem vessels, thin-walled fibres, slower growth rates and small tree sizes.