1 Microstructural characterization of commercial kiwifruit 2 cultivars using X-ray micro computed tomography 3 Dennis Cantrea, Andrew Eastb, Pieter Verbovena, Ximenita Trejo Arayab, Els Herremansa, 4 Bart M. Nicolaïa, Thamarath Pranamornkithb, Michael Lohc, Alistair Mowatd, Julian Heyesb. a 5 6 b BIOSYST-MeBioS, KU Leuven, Willem de Croylaan 42, Heverlee, Belgium Centre for Postharvest and Refrigeration Research, Massey University, Palmerston North, 7 8 9 New Zealand c Fonterra Research and Development Centre, Palmerston North, New Zealand d Zespri™ International Ltd, Mt Maunganui, New Zealand 10 11 Corresponding author: 12 Andrew East, Centre for Postharvest and Refrigeration Research, Massey University, 13 Palmerston North, New Zealand 14 Tel.: +64 350 9099; e-mail: A.R.East@massey.ac.nz 15 16 1 ABSTRACT 17 The skin is the physical barrier between the fruit and the environment in which it develops. 18 Environmental conditions during fruit development have a large influence on fruit quality, 19 both at the time of harvest and during subsequent storage. It is hypothesized that some 20 features of the skin and sub-epidermal tissues could provide information about the past 21 growing conditions to which the fruit was exposed and therefore be of predictive value for 22 storage quality. In this study, five commercial kiwifruit cultivars (‘Hayward’, ‘Hort16A’, 23 ‘G3’, ‘G9’ and ‘G14’) were studied, and ‘Hayward’ fruit were manipulated during growth 24 with different light or cultural practices. After harvest at horticultural maturity, X-ray micro 25 computed tomography (µCT) was used to investigate features of the skin and the immediate 26 parenchyma tissue. Despite orchard management practices (crop load and girdling) being 27 observed to effect macro fruit quality parameters (mass, firmness, SSC, and DM), differences 28 in microstructure (e.g. porosity) caused by these practices were not observed. However, 29 porosity and pore size were found to be highly variable between cultivars. The thickness of 30 dense sub-epidermal tissue could be readily measured and the 3-D distribution of raphide 31 bundles was visible as high density particles distributed within the parenchyma. Overall, 32 µCT was found to be a powerful technique to explore fruit epidermal and sub-epidermal 33 structures in three dimensions at a micro level. However, the length of time required for data 34 capture and analysis and the large number of samples required to overcome natural variation 35 within horticultural products need to be considered. Future work may define the impact of 36 differences in porosity or sub-epidermal anatomy on kiwifruit physiology (e.g. firmness 37 change or sensitivity to low oxygen storage atmospheres). With this information, µCT could 2 38 be used as a screening tool during plant breeding, or to determine the response to agronomic 39 treatments, without conducting lengthy storage trials. 40 41 Highlights 42 Microstructural differences were found among kiwifruit cultivars 43 Differences in raphide were found among kiwifruit cultivars 44 Orchard manipulation caused macro quality differences but not microstructural 45 46 47 Keywords: X-ray µCT, microstructure, cultivar differences, orchard manipulation, raphide 48 49 2 Introduction Kiwifruit quality and storability are known to be affected by environmental factors 50 during growth and development. Sunlight exposure during development can enhance 51 pigment development, stimulate fruit maturation, enhance skin coloration and result in firmer 52 fruit with less incidence of rot in storage (Tombesi et al., 1993) and better fruit storability 53 (Antognozzi et al., 1995). This sunlight exposure is directly influenced by canopy density 54 (Snelgar et al., 1998). Similarly, application of trunk girdling has been applied to result in 55 both increased fruit mass and dry matter, depending on timing, while fruit crop load is known 56 to affect final fruit size (Patterson and Currie, 2011). Further studies in manipulating 57 appropriate growth conditions may provide an insight in alleviating LTB in kiwifruit during 58 storage. It is hypothesized that membrane deterioration plays a critical role during ripening 59 and LTB development (Marangoni et al., 1996). Membrane deterioration affects porosity 60 either through filling of pores with cellular fluid as a result of cellular breakdown (Kuroki et 61 al., 2004), or creation of larger air spaces as a result of tissue collapse (Herremans et al., 62 2013a). This membrane deterioration leading to changes in porosity is also suggested to play 63 a critical role in chilling injury in mangoes. Chilling injury damages cell wall causing cell 64 wall disassembly and cell leakage (Han et al., 2006). Thus, cellular fluid may fill the 65 intercellular gas-filled spaces, completely separating the pores (Narain et al., 1998). The 66 decrease in pore size and pore connectivity in cucumber during ripening also suggests 67 leakage of intracellular substance into the gas-filled intercellular space (Kuroki et al., 2004). 68 The importance of pore network and void volume fraction for gas exchange in apples 69 and pears is extensively studied (Herremans et al., 2013a,b; Ho et al., 2013a, b; Verboven et 4 70 al., 2008). For example, some cultivars (e.g., ‘Braeburn’ and ‘Cripps Pink’) develop internal 71 browning disorders in storage (Herremans et al., 2013a; James and Jobling, 2009), while 72 others (e.g. ‘Granny Smith’) tend to develop skin related injuries (Fan et al., 1999). 73 Contemporary research suggests that these differences in storability and susceptibility to 74 disorders are associated with skin and flesh properties that influence gas diffusion within the 75 fruit tissue (Ho et al., 2009, 2010, and 2011) including differences in porous microstructure 76 (Ho et al., 2013b), with those differences strongly affecting optimal storage conditions (Ho et 77 al., 2013c). Whether such relationships of flesh properties to optimal storage conditions exist 78 for kiwifruit is currently unknown. 79 Techniques to visualise and quantify microstructure have been used to answer valid 80 research questions and to provide a better understanding on the relation of microstructure to 81 quality characteristics during storage. Unlike light and electron microscopy, non-destructive 82 techniques provide a more accurate depiction of the microstructure since cells and tissues 83 remains intact and undamaged and no tedious sample preparation is required (Veraverbeke et 84 al., 2001; Musse et al., 2010). One such non-destructive technique is X-ray micro computed 85 tomography (μCT), a three-dimensional visualization technique that creates an image based 86 on X-ray attenuation within the sample. The difference in X-ray attenuation of different 87 materials creates contrast to differentiate low density and high density materials 88 (Karathanasis and Hajek, 1996). In comparison to other non-destructive techniques, µCT has 89 excellent spatial resolution that resolves intercellular space down to submicron range, and 90 depth of penetration allowing subsurface features to be studied. This technique can help to 91 characterize and understand foods by measuring cell size and shape, void space and spatial 92 distribution (Lim and Barigou, 2004). The use of µCT allowed visualisation of the fruit void 93 network architecture, showing a volume fraction of 5.1 % and 23 % for ‘Conference’ pear 94 and ‘Jonagold’ apple cortex, respectively (Verboven et al., 2008; Mendoza et al., 2007). This 95 technique was also successfully used in 3-D analysis of raphides in rose peduncles and Lotus 96 miyakojimae seeds (Matsushima et al., 2012; Yamauchi et al., 2013). 97 The main objective of this study was to gain the first insights into the potential 98 differences in 3-D microstructural properties of commercial kiwifruit genotypes. μCT was 99 used as a tool to explore the fruit peel and tissue from five cultivars. Additionally, treatments 100 to deliberately manipulate ‘Hayward’ kiwifruit during growth via altering light conditions, 101 crop load and application of girdling were applied as a preliminary study to investigate if any 102 of the crop manipulation techniques would influence cellular arrangement and density within 103 the fruit. Insights from this study are expected to reveal relationship between the potential for 104 storage capability of cultivars to the changes in cellular arrangement and density as affected 105 by environmental factors during growth and development. 106 3 Materials and methods 107 3.1 Plant material and treatment manipulation 108 Currently there are five commercial cultivars exported from New Zealand, being green 109 cultivars ‘Hayward’ (A. deliciosa, Zespri™ Green) and ‘G14’ (A. deliciosa × chinensis, 110 Zespri™ Sweet Green) and yellow fleshed cultivars (all A. chinensis) ‘Hort16A’ (Zespri™ 111 Gold), ‘G3’ (Zespri™ Sun Gold) and ‘G9’ (Zespri™ Charm). Single tray samples of ‘G14’, 112 ‘Hort16A’, ‘G3’, and ‘G9’ were obtained from commercial orchards in New Zealand and 113 delivered via airfreight to KU Leuven, Belgium, in June 2013. Five fruit from each cultivar 114 were scanned for X-ray µCT. At the time of the measurements, fruit could be considered as 115 eating ripe. 6 116 In a growing condition manipulation experiment conducted in the 2013 harvest season, 117 a mature ‘Hayward’ kiwifruit orchard trained on a pergola system and located in Te Puna, 118 Bay of Plenty, New Zealand was used. This experiment consisted of a 2 × 2 matrix of plant 119 manipulation treatments, being fruit crop load (36 or 43 t/ha) and usage (or not) of trunk 120 girdling (Patterson and Currie, 2011). Crop thinning occurred on 4 - 5 Jan 2013 and trunk 121 girdling occurred on 10 Dec 2012 and 2 Feb 2013. A single sample of 15 fruit was harvested 122 from each of the 4 treatments on 7 May 2013 and delivered by airfreight to KU Leuven. Five 123 fruit from each treatment were measured in June at which time the fruit could be considered 124 eating ripe. Commercial harvest of this experiment occurred on 15 May 2013 at which time 125 90 fruit samples were collected and subsequently analysed for at harvest quality (mass, 126 soluble solids content, firmness and % dry matter). 127 128 3.2 Micro X-ray CT imaging 129 Kiwifruit samples measuring 5 mm x 5 mm x 10 mm including the skin were excised 130 from the equatorial region and wrapped in parafilm to prevent dehydration prior to 131 subsequent CT scanning. A Skyscan 1172 high resolution µCT scanner (Bruker, Kontich, 132 Belgium) was used for acquiring projection images with a source power of 10 W at 60 kV 133 and 167 µA. Each projection image was averaged from three frames with each frame taken 134 with an exposure time of 295 ms. The sample was rotated on the stage at an increment of 0.4° 135 until a rotation angle of 204.8° (180° + fan beam angle) was completed generating 512 136 shadow projections with a pixel size of 4.87 µm. Cross section slices were generated from 137 the shadow projections using the Feldkamp reconstruction algorithm (Feldkamp et al., 1984) 138 implemented in Nrecon 1.6.5.8 software (http://www.skyscan.be/next/downloads.htm). 139 Reconstruction parameters for beam hardening, ring artefact reduction and smoothing were 140 set to 35, 8 and 2, respectively. The dynamic range or linear attenuation range was limited to 141 0 – 0.0854 to generate an 8-bit bitmap grayscale cross section slice. 142 3.3 Image processing 143 Undamaged and intact µCT tomographs measuring 2 mm x 2 mm x 2 mm, left after 144 virtual cropping, were utilised for 3-D image processing and analysis (Figure 1A). Each 145 image was segmented with a manual threshold (Figure 1B) to obtain the pores and cell 146 assembly, followed by individual pore labelling (Figure 1C) and 3-D image rendering of the 147 pore network (Figure 1D, Figure 2B). A spatial graph representation of the skeleton using 148 TEASAR algorithm (Sato et al., 2000) was made to show the essential geometry and local 149 thickness of the pore network (Figure 2C). Microstructural parameters (Table 1) as described 150 by Herremans et al. (2013a) were analysed. Raphide bundles were revealed as a substantial 151 number of high density oblate spheroid particles in the epidermal and sub-epidermal region. 152 These particles were segmented with a manual threshold and labelled prior to surface 153 rendering and subsequent 3-D analysis. Image processing and 3-D analysis were performed 154 using CTAn 1.12 (Bruker microCT, Kontich, Belgium) and Avizo 7.1 (VSG, Bordeaux, 155 France). Results of the quality measurements and microstructural analysis were subjected to 156 analysis of variance using PROC GLM procedure in SAS 9.3 (SAS Institute, Cary, NC, 157 USA). 158 3.4 Fruit quality measurement 159 At harvest, fruit quality measures from the production method manipulation experiment 160 applied to ‘Hayward’ kiwifruit were conducted in order to demonstrate the potential effects 161 of these treatments on fruit quality. Individual fruit mass (g) was measured using a balance 8 162 (PG-503S, Mettler-Toledo GmbH, Greifensee, Switzerland) with 0.001 g accuracy. Fruit dry 163 matter content was determined using oven drying techniques. An equatorial fruit slice with a 164 thickness of 2-3 mm thick and of known mass was dried at 60-65°C for 24 h, subsequently 165 weighed and dry matter expressed as a percentage of the wet mass. Firmness was measured 166 using a QALink Penetrometer (Willowbank Electronics Ltd., Napier, New Zealand), fitted 167 with the standard 7.9 mm round Effegi probe. Prior to measurement, 1 mm of the fruit skin 168 was removed at two locations (90° apart) around the equator of the fruit with a fruit slicer 169 (setting depth: 1 mm). The average peak force (N) required to puncture the skinned tissue to a 170 depth of 8 mm at a speed of 20 mm.s-1 was recorded. Soluble solids were measured as °Brix, 171 using a digital pocket refractometer (PAL-1, Atago, Japan). Juice was taken from both the 172 stem and blossom ends of the fruit and mixed in approximately equal amounts on the 173 refractometer. 174 4 175 4.1 Results Imaging and rendering 176 The microtomographs of kiwifruit tissue from five commercial cultivars showed 177 differences in grey scale values according to the X-ray attenuation by the different elements 178 of the tissue (Figure 2A). The shade of the cells is lighter than that of the pores as a 179 consequence of the higher density of the cells compared to that of air. 3-D renderings of pore 180 structure and pore network model of samples are shown in Figure 2B and 2C respectively. 181 Both figures seem to indicate considerable differences in tissue microstructure between fruit 182 of different cultivars. 183 4.2 Differences between cultivars 184 3-D rendering of pore structure and pore network model (Figure 2B and 2C) suggests the 185 presence of isolated, short and narrow pores in ‘Hayward’ and ‘G3’ while ‘Hort16A’, ‘G9’ 186 and ‘G14’ showed highly branched, more connected, and thicker pores. The porosity seems 187 to be low in ‘G3’ and high in‘Hort16A’ and ‘G14’ cultivars. The statistical analysis of 188 microstructural features (Table 2) emphasized cultivar specific difference in porosity of 189 kiwifruit. ‘G14’ and ‘Hort16A’ had a significantly higher porosity compared to ‘Hayward’ 190 and‘G3’ cultivars (Table 2). In addition, fruit of 'G3' had porosity value significantly lower 191 than those of the other cultivars. Pores observed below the skin (Figure 2) also differ among 192 cultivars and they may significantly affect the gas exchange. ‘G3’ cultivar, for example, 193 showed larger pores beneath the skin and smaller pores in the parenchyma tissue. All the 194 other cultivars have small isolated pores under the skin but have large and more connected 195 pores located in the parenchyma tissue. 196 The size range of pores varied by an order of magnitude of 4 from the detection limit of 197 2.99 × 10-6 mm3 (Herremans et al., 2013a) to pore volumes as high as 1.35 × 10-2 mm3. The 198 pore size distribution (Figure 3) showed further differences between kiwifruit cultivars. ‘G14’ 199 and ‘Hort16A’ showed the broadest pore size distribution with a pore equivalent diameter up 200 to 295 µm. 3-D microstructural analysis data (Table 2) also showed that ‘G14’ had the 201 highest maximal individual pore volume. The prevalence of large pores (> 100 µm) in ‘G14’, 202 contributing to 40% of its total pore volume, also separates it from other cultivars. In contrast, 203 ‘G3’ had a narrow pore size range with only 196 µm as the largest pore equivalent diameter 204 detected, and with only 0.0024 mm3 maximal individual pore volume, the lowest of the five 205 cultivars. It is also differentiated by the prevalence of small pores (< 50 µm) which constitute 206 58% of its total pore volume, significantly higher across all cultivars analysed. Pore sizes of 10 207 ‘G9’, ‘Hort16A’, and ‘Hayward’ appeared to fall in a similar and intermediate range. Small 208 pores (< 50 µm) represent a substantial portion (more than 90%) of the total number of pores 209 in all cultivars. 210 Pore structure thickness results concur with observations on pore size. The large pores 211 in ‘G14’ are accompanied by a thick pore structure which is significantly higher than all the 212 other cultivars (Table 2). In comparison, the pores found in ‘G3’ cultivar have a significantly 213 lower pore structure thickness compared to other cultivars due to its narrow pores. Pore 214 structure thickness of ‘G9’, ‘Hort16A’, and ‘Hayward’ appeared to fall in an intermediate 215 range. The pore specific surface area was lowest in ‘G14’ and highest in ‘G3’ (Table 2). The 216 smaller pore size fraction in ‘G3’ provided higher surface area per unit volume of tissue 217 giving a higher value of pore specific surface area. 218 Pore connectivity and pore fragmentation results (Table 2) confirmed the observation 219 on the pore network model suggesting highly fragmented and less connected pores in ‘G3’. 220 Furthermore, the pore fragmentation and pore connectivity of ‘G3’is significantly different 221 across all cultivars studied. The observed highly connected pores in ‘G14’ from the pore 222 network model correspond to its low pore fragmentation value. In addition, the ‘G14’ pore 223 fragmentation is significantly lower than that of all other cultivars. Pore connectivity results, 224 on the other hand, showed no significant difference between ‘G14’, ‘G9’, ‘Hort16A’, and 225 ‘Hayward’, suggesting the same number of connected structures per unit volume of tissue in 226 these cultivars. 227 The peel thickness, defined as the distance between the fruit surface and the first pore, 228 was not significantly different amongst the 5 cultivars (Table 2). The peel thickness was on 229 average between 70 and 85 µm. 230 4.3 Effect of growing conditions 231 As expected, manipulation of growing conditions resulted in differences in ‘Hayward’ 232 fruit quality attributes at commercial harvest (Table 3 and 4). High crop load resulted in 233 reduced fruit size (mass) and soluble solids content. High crop load also reduced dry matter 234 when the vines were girdled, and increased fruit firmness when the vines were not girdled. 235 Trunk girdling caused increased fruit size (mass), soluble solids content and decreased 236 firmness. However, X-ray CT image analysis failed to identify any significant difference in 237 equivalent porosity (Table 4), pore diameter, porosity or peel thickness effects (data not 238 shown) caused by either pre-harvest crop load or trunk girdling treatments. 239 240 . 4.4 Presence of raphides 241 A substantial number of oblate spheroid shaped high density particles were observed 242 within the kiwifruit tissue. These were observed distinctly on the X-ray images (Figure 4). 243 These particles were approximately 80-130 µm in length (Figure 5) and in most occasions 244 orientated with the long axis parallel with the kiwifruit skin surface. The particles were 245 widely spread across the whole pericarp and beneath the surface of the skin. All five 246 kiwifruit cultivars showed the presence of these particles, identified as druses (bundles of 247 raphides). High counts of these particles were found in ‘G14’ and ‘G3’ (Figure 5) but fruit to 248 fruit variability rendered no significant difference in count of these bundles across all 249 cultivars. ‘G3’ had the highest mean raphide bundle volume and length. It also contained 250 significantly larger raphide bundles compared to ‘Hort16A’ and ‘G14’. More spherical 251 bundles were found in ‘G14’ and ‘Hort16A’ cultivars than in ‘Hayward’. 12 252 5 Discussion 253 The porosity and pore sizes of kiwifruit are generally small compared to other fruit. For 254 example, for ‘Braeburn’ apple, Herremans et al. (2013a) found pore sizes up to 600 µm with 255 50% of the pores smaller than 180 µm, thus considerably larger than those of kiwifruit. This 256 is not unexpected because apple consists also of larger cells, and has a much larger porosity 257 with values up to 25.4% compared to kiwifruit with values up to 6.5%. The porosity and 258 small pore size of kiwifruit are more similar to pear and cucumber fruit with values up to 4.1 259 and 3.4% respectively (Verboven et al., 2008; Kuroki et al., 2004). We earlier attributed these 260 differences to differences in the mechanisms of pore formation where programmed cell death 261 could be responsible for larger (lysogenous) pores in apple and cell wall separation for 262 smaller (schizogenous) pores in pears. More likely the latter mechanisms is responsible for 263 pores in kiwifruit as well. In terms of the number of pores, kiwifruit is also similar to 264 cucumber: most (79%) of the pores were smaller than 26 µm equivalent diameter, with 20% 265 of the pores were between 27 and 58 µm and fewer than 1% were larger. Kiwifruit had a 266 somewhat wider size range distribution with approximately 65% of the pores smaller than 20 267 µm, 30% for pores between 20 and 60 µm and 5% larger than 60 µm. Concerning the range 268 of porosity values found for kiwifruit cultivars, the least porous cultivar ‘G3’ has a porosity 269 which is only 25% of that of the most porous fruit (‘G14’). For comparison, analysis of 3 270 apple cultivars and ‘Conference’ pear showed that ‘Kanzi’ had a porosity that was 271 approximately 50% that of ‘Jonagold’, while conference pear had a porosity that was less 272 than 20% of that of ‘Jonagold’ (Herremans et al. 2013b). The porosity of kiwifruit ranging 273 from 1.55 % in ‘G3’ up to 6.5 % in ‘G14’ is comparable to that of ‘Conference’ pear 274 (Herremans et al., 2013b) and that of the radial pericarp of tomato (Musse et al., 2010) with 275 porosity values of 4.1 and 3.0 % respectively. 276 Comparing the pore network of kiwifruit and ‘Jonagold’ apples, kiwifruit has pore 277 structure thickness of that is only 28-40 % with that of ‘Jonagold’. Pore fragmentation on the 278 other hand, is more than twice that of ‘Jonagold’ (Herremans et al., 2013a). Physically, fruit 279 with larger, but especially more connected, air voids enable more efficient gas exchange (Ho 280 et al., 2013b). 281 diffusivity and, hence, sensitivity to controlled atmosphere conditions. Physically, more air 282 voids may affect the cell structure and mechanical properties of the tissue and how this 283 weakens with pectin degradation during ripening and storage (Waldron et al., 1997). These differences in pore structure may account for differences in gas 284 285 Future scans at a higher resolution may reveal pores that are smaller than the detection 286 limit. These small pores may further contribute to the pore connectivity and pore analysis 287 result based on the current pixel resolution. Further, the hypothetical presence of small pores 288 would be more important in less porous kiwifruit cultivars where they would significantly 289 contribute to gas exchange 290 291 Kiwifruit peel thickness falls within the range that was measured for apple 292 (Veraverbeke et al., 2001; Verboven et al., 2013). Apple has a peel thickness between 60 and 293 120 µm, depending on the cultivar, with significant differences between ‘Idared’, on the one 294 hand, and ‘Arlet’, ‘Braeburn’ and ‘Royal Gala’, on the other hand. The thickness of the apple 295 wax layer also varies depending on the cultivar, with ‘Elstar’, ‘Jonagored’ and ‘Jonagold’ 296 having a wax layer thickness significantly different from each other (Veraverbeke et al., 14 297 2001). These large cultivar differences in porosity and sub-epidermal anatomy emphasise the 298 value of a technique such as µCT as a screening tool in a breeding programme, once a link 299 between these features and fruit storage performance is established. 300 301 This work used commercially-produced fruit which were measured at a table-ripe 302 stage. Micro X-ray CT studies suggest that apple browning is associated with cellular 303 breakdown which was detected as a disappearance of air spaces in affected tissue followed by 304 the creation of large pores after water has diffused out of the affected zone (Herremans et al., 305 2013a). In the case of kiwifruit, it may be possible that as kiwifruit develop LTB or ripen, 306 they may also go through porosity changes throughout storage. In investigating kiwifruit 307 ripening with MRI, Taglienti et al. (2009) suggested that as kiwifruit softened, water was “re- 308 organised” due to a “release” of water from the cellular matrix. Changes in porosity were 309 also observed during ripening of mango. In addition, microstructural changes observed 310 during mango ripening also suggest a release of water from the cellular matrix (Cantre et al., 311 2013). Use of X-ray µCT to determine porosity changes (if any) during ripening and/or LTB 312 development may assist in developing understanding of these processes. 313 A number of plant species including kiwifruit are known to contain druses. These are 314 bundles of raphides or calcium oxalate crystals which promote the burning or ‘catching’ 315 sensation in the mouth when eaten due to their needle-like nature (Perera et al., 1990). 316 Normally, druses are enclosed inside an idioblast cell in the locular gel around the seeds 317 (Perera et al., 1990). Several roles are proposed for raphides: protection against animals, 318 detoxifying oxalates, contributing to tissue firmness and regulating free calcium 319 concentration in the fruit (Faheed et al., 2012; Nakata, 2012). Calcium uptake and deposition 320 during development depends on environmental conditions (Buxton, 2005). The distribution, 321 number and morphological characteristics of these particles were not known prior to this 322 study. 323 Raphides in rose peduncles and Lotus miyakojimae seeds were successfully quantified 324 in 3-D using synchrotron X-ray µCT (Matsushima et al., 2012; Yamauchi et al., 2013), but 325 the ease of detection of raphides by conventional X-ray µCT has not been previously 326 reported. Furthermore, the presence of raphides this close to the skin is not normally reported 327 or visible with conventional staining systems (e.g., Hallett et al., 2005). This holds promise 328 for studies of their morphological characteristics and 3-D arrangement in space utilizing only 329 a laboratory scale X-ray µCT system. Threshold-binarisation of the µCT images facilitated 330 isolation and labelling of these particles before subsequent 3-D analysis and quantification. 331 X-ray µCT further allowed determination of the amount and morphological characteristics of 332 raphides for each kiwifruit cultivar. The non-invasive nature of X-ray µCT provides an ideal 333 tool for in-situ investigation of raphides and will facilitate studies of changes in calcium 334 oxalate deposition. If these raphides act as a source or competes for calcium available for 335 cross-linking pectins in the apoplast of soft fruit (Perera et al., 1990), the observed cultivar 336 differences in size and shape may become important in influencing firmness. 337 6 338 Conclusions This work demonstrates X-ray µCT as a research technique to investigate kiwifruit 339 tissue in detail, revealing its pore structure, porosity and raphide features. X-ray µCT 340 revealed the microstructure and showed differences in the cell/void networks near the skin of 341 5 different cultivars. Yellow fleshed ‘G3’ was found to have a lower porosity due to a 16 342 majority of small pores, whereas ‘G14’ had the greatest porosity as a result of a high number 343 of larger voids. Other cultivars were intermediary. Orchard management practices (e.g. 344 girdling and crop loading) used to effect fruit quality parameters (mass, firmness, SSC, and 345 DM) did not show any observable differences in microstructure. However, other growth 346 conditions and phenotypic variation may have an effect on porosity and microstructure and 347 this remains to be investigated. Raphides were also identified and classified very close to the 348 fruit surface with cultivar differences in size and shape characterised, which may influence 349 calcium relations (and firmness) in these cultivars. 350 Acknowledgements 351 The work is an output of the Food Structure Design Theme of the Fonterra / Zespri™ 352 Primary Growth Partnership research project funded by the Ministry of Primary Industries, 353 New Zealand. The authors would like to acknowledge the wider staff and student group of 354 the Centre for Postharvest and Refrigeration Research, Massey University, and PlusGroup™ 355 Horticulture who donated their time to enable experimental set-up and fruit harvest. Dennis 356 Cantre is an IRO scholar of KU Leuven. Els Herremans is a doctoral fellow of the IWT 357 (Flemish agency for Innovation by Science and Technology). 358 7 359 Antognozzi, E., Boco, M., Famiani, F., Palliotti, A., Tombesi, A., 1995. Effect of different 360 light intensity on quality and storage life of kiwifruit. Acta Horticulturae 379, 483-490. References 361 362 Buxton, K. N., 2005. Preharvest practices affecting postharvest quality of 'Hayward' 363 kiwifruit. Doctor of Philosophy Thesis, Massey University, Palmerston North. 364 365 Cantre, D., Herremans, E., Verboven, P., Ampofo-Asiama J. Nicolai, B., 2013. 366 Characterization of the 3-D microstructure of mango (Mangifera indica L. cv. Carabao) 367 during ripening using X-ray computed microtomography. Innov. Food Sci. Emerg. 368 doi:10.1016/j.ifset.2013.12.008 369 370 Faheed, F., Mazen, A., Elmohsen, S., 2012. Physiological and ultrastructural studies on 371 calcium oxalate crystals formation in some plants. Turk. J. Botany 37, 139-152. 372 373 Fan, X., Mattheis, J. P., Blankenship, S., 1999. Development of superficial scald, soft scald, 374 core flush and greasiness is reduced by MCP. J. Agric. Food Chem. 47, 3063-3068. 375 376 Feldkamp, L. A., Davis, L. C., Kress, J. W., 1984. Practical cone-beam algorithm. J. Op. Soc. 377 Am. A.1(6), 612–619. 378 379 Hallett, I. C., Sutherland, P.W., Harker, F. R., and MacRae, E. A. 2005. Fruit cell walls: 380 Texture and convenience. Proc. Microscopy and Microanalysis 11, 1144-1145. 381 382 Hahn, M., Vogel, M., Pompesius-Kempa, M., Delling, G., 1992. Trabecular bone pattern 383 factor – a new parameter for simple quantification of bone microarchitecture. Bone 13, 327– 384 330. 385 18 386 Han J, Tian S, Meng X, Ding Z, 2006. Response of physiologic metabolism and cell 387 structures in mango fruit to exogenous methyl salicylate under low-temperature stress. 388 Physiologia Plantarum. 128, 125-133. 389 390 Herremans, E., Verboven, P., Bongaers, E., Estrade, P., Verlinden, B.E., Wevers, M., Hertog, 391 M., Nicolai, B., 2013a. Characterisation of ‘Braeburn’ browning disorder by means of X-ray 392 micro-CT. Postharvest Biol. Technol. 75, 114-124. 393 394 Herremans, E., Verboven, P., Bongaers, E., Estrade, P., Verlinden, B., Wevers, M., Nicolai, 395 B., 2013b. Isolation of single cells and pores for the characterisation 3-D fruit tissue 396 microstructure based on X-ray micro-CT image analysis. Submitted. 397 398 Hildebrand, T., Ruegsegger, P., 1997. A new method for the model independent assessment 399 of thickness in three dimensional images. J. Microscopy 185, 67–75. 400 401 Ho, Q. T., Verboven, P., Mebatsion, H. K., Verlinden, B. E., Vandewalle, S., Nicolaï, B. M., 402 2009. Microscale mechanisms of gas exchange in fruit tissue. New Phytol. 182, 163-174. 403 404 Ho, Q. T., Verboven, P., Verlinden, B., Schenk, A., Delele, M., Rolletschek, H., Vercammen, 405 J., Nicolai, B., 2010. Genotype effects on internal gas gradients in apple fruit. J. Exp. Bot. 406 61(10), 2745-2755. 407 408 Ho, Q. T., Verboven, P., Verlinden, B., Herremans, E., Wevers, M., Carmeliet, J., Nicolai, B., 409 2011. A 3-D multiscale model for gas exchange in fruit. Plant Physiol. 155(3), 1158-1168. 410 411 Ho, Q., Carmeliet, J., Datta, A., Defraeye, T., Delele, M., Herremans, E., Opara, L., Ramon, 412 H., Tijskens, E., Van der Sman, R., Van Liedekerke, P., Verboven, P., Nicolai, B., 2013a. 413 Multiscale modeling in food engineering. J. Food Eng. 114(3), 279-291. 414 415 Ho, Q., Verboven, P., Fanta, S., Abera, M., Retta, M., Herremans, E., Defraeye, T., Nicolai, 416 B., 2013b. A multiphase pore scale network model of gas exchange in apple fruit. Food 417 Bioprocess Technol. doi:10.1007/s11947-012-1043-y 418 419 Ho, Q., Verboven, P., Verlinden, B., Schenk, A., Nicolai, B., 2013c. Controlled atmosphere 420 storage may lead to local ATP deficiency in apple. Postharvest Biol. Technol. 78, 103-112. 421 422 James, H. J., Jobling, J. J., 2009. Contrasting the structure and morphology of the radial and 423 diffuse flesh browning disorders and CO2 injury of ‘Cripps Pink’ apples. Postharvest Biol. 424 Technol. 53, 36-42. 425 426 Jennings, B. R., Parslow, K., 1988. Particle size measurement: the equivalent spherical 427 diameter. Proceedings of the Royal Society of London Series A 419, 137-149 428 429 Karathanasis, A.D., Hajek, B.F., 1996. Elemental analysis by X-ray fluorescence 430 spectroscopy, in: Sparks, D.L., (Ed.), Methods of Soil Analysis. Soil Science Society of 431 America, Madison, WI, pp. 161-223. 432 20 433 Kuroki, S., Oshita S., Sotome, I., Kawagoe, Y., Seo, Y., 2004. Visualization of 3-D network 434 of gas-filled intercellular spaces in cucumber fruit after harvest. Postharvest Biol. Technol. 435 33, 255-262. 436 437 Lim, K. S., Barigou, M., 2004. X-ray micro-computed tomography of cellular food products. 438 Food Res. Int. 37, 1001-1012. 439 440 Marangoni, A.G., Palma, T., Stanley, D.W., 1996. Membrane effects in postharvest 441 physiology. Postharvest Biol. Technol. 7, 193-217. 442 443 Matsushima, U., Hilger, A., Graf, W., Zabler, S., Manke, I., Dawson, M., Choinka, G., 444 Herppinch, W., 2012. Calcium oxalate crystal distribution in rose peduncles: Non-invasive 445 analysis by synchrotron X-ray micro-tomography. Postharvest Biol. Technol. 72, 27-34. 446 447 Mendoza, F., Verboven, P., Mebatsion, H., Kerckhofs, G., Wevers, M., Nicolaï, B., 2007. 448 Three-dimensional pore space quantification of apple tissue using X-ray computed 449 microtomography. Planta 226, 559-570. 450 451 Musse, M., De Guio, F., Quellec, S., Cambert, M., Challois, S., Davenel, A., 2010. 452 Quantification of microporosity in fruit by MRI at various magnetic fields: comparison with 453 X-ray microtomography. Magn. Reson. Im. 28, 1525-1534. 454 455 Nakata, P. A., 2012. Plant calcium oxalate crystal formation, function, and its impact on 456 human health. Front. Biol. 7, 254-266. 457 458 Narain, N., Bora, P. S., Narain, R., Shaw, P.E., 1998. Mango, in: Shaw, P.E., Chan, H., Nagy, 459 S. (Eds.), Tropical and Subtropical Fruits. Agscience Inc., Aubumdale, Fla., USA, pp. 1-77. 460 461 Odgaard, A., Gundersen, H.J., 1993. Quantification of connectivity in cancellous bone, with 462 special emphasis on 3-D reconstructions. Bone 14, 173-182. 463 464 Patterson, K. J., Currie, M. B., 2011. Optimising kiwifruit vine performance for high 465 productivity and superior fruit taste, Acta Horticulturae 913, 257-268. 466 467 Perera, C. O., Hallett, I. C., Nguyen, T. T., Charles. J.C., 1990. Calcium oxalate crystals: 468 The irritant factor in kiwifruit. J. Food Sci. 55, 1066-1070. 469 470 Sato, M., Bitter, M., Bender, M.A., Kaufman, A.E., Nakajima, M., 2000. TEASAR: Tree- 471 structure Extraction Algorithm for Accurate and Robust Skeletons. The Eighth Pacific 472 Conference on Computer Graphics and Applications Proceedings, 281-287. 473 474 Snelgar, W.P., Hopkirk, G., Seelye, R.J., Martin, P.J., Manson, P.J., 1998. Relationship 475 between canopy density and fruit quality of kiwifruit. N. Z. J. Crop. Hortic. Sci. 26, 223-232. 476 477 Taglienti, A., Massantini, R., Botondi, R., Mencarelli, F., Valentini, M., 2009. Postharvest 478 structural changes of ‘Hayward’ kiwifruit by means of magnetic resonance imaging 479 spectroscopy, Food Chem. 114, 1583-1589. 480 22 481 Tombesi, A., Antognozzi, E., Palliotti, A., 1993. Influence of light exposure on 482 characteristics and storage life of kiwifruit. N. Z. J. Crop Hortic. Sci. 21, 87-92. 483 484 Veraverbeke, E., Van Bruaene, N., Van Oostveldt, P., Nicolai, B., 2001. Non-destructive 485 analysis of the wax layer of apple (Malus domestica Borkh.) by means of confocal laser 486 scanning microscopy. Planta 213, 525-533. 487 488 Verboven, P., Kerckhofs, G., Mebatsion, H. K., Ho, Q. T., Temst, K., Wevers, M., et al., 489 2008. Three-dimensional gas exchange pathways in pome fruit characterized by synchrotron 490 X-ray computed tomography. Plant Physiol. 147, 518-527. 491 492 Verboven, P., Nemeth, A., Abera, M., Bongaers, E., Daelemans, D., Estrade, P., Herremans, 493 E., Hertog, M., Saeys, W., Vanstreels, E., Verlinden, B., Leitner, M., Nicolai, B., 2013. 494 Optical coherence tomography visualizes microstructure of apple peel. Postharvest Biol. 495 Technol. 78, 123-132. 496 497 Waldron, K. W., Smith, A. C., Parr, A. J., Ng, A., Parker, M., 1997. New approaches to 498 understanding and controlling cell separation in relation to fruit and vegetable texture. Trends 499 Food Sci. Technol. 8, 213-221. 500 501 Yamauchi, D., Tamaoki, D., Hayami, M., Takeuchi, M., Karahara, I., Sato, M., Tayooka, K., 502 Nishioka, H., Terada, Y., Uesugi, K., Takano, H., Kagoshima, Y., Mineyuki, Y., 2013. 503 Micro-CT observations of the 3-D distribution of calcium oxalate crystals in cotyledons 504 during maturation and germination in Lotus miyakojimae seeds. Microscopy, 62, 353-361. 505 506 8 Tables 507 Table 1. Morphometric parameters and description used to quantify microstructure. Microstructural Unit Description Volume mm3 Volume of the sample Area mm2 Area of the sample boundary Specific surface area mm-1 The surface area of all the solid objects in the sample parameter divided by the volume of the object Porosity % Pore volume divided by total volume of the analysed sample (pores and cells) Connectivity (-) Defined as a measure of the degree to which a structure is multiply connected. For a network, it represents the maximal number of branches that can be broken before the structure is separated into two parts (Odgaard and Gunderson, 1993). Pore equivalent diameter µm The diameter of a sphere of equivalent volume as the irregularly shaped object (Jennings and Parslow, 1988). Pore structure thickness mm Average of the local thickness of the pores. Calculated by skeletonisation of the binarised tissue, followed by a sphere fitting algorithm for each voxel of the skeleton (Hildebrand and Ruegsegger, 1997). Fragmentation index mm-1 Inverse index of connectivity, calculated by comparing 24 volume and surface of the binarised object before and after an image dilation (Hahn et al., 1992). A lower fragmentation signifies better connected structures. 508 Herremans et al. (2013a) 509 Table 2. Microstructural parameters of pores and tissue, statistics of the pore size distribution, and peel thickness at eating ripe condition. Mean 510 values are presented with their 95% confidence interval. Values were averaged from 5 fruits per cultivar except for porosity and peel thickness 511 of ‘Hayward’ where values were averaged from 20 fruits. Means followed by different letters are significantly different at 0.05 level of 512 significance. Cultivar G9 Hort16A G3 G14 Hayward 4.6 ± 1.2 (AB) 5.6 ± 1.2 (A) 1.55 ± 0.43 (C) 6.49 ± 0.75 (A) 3.94 ± 0.40 (B) 0.0062 ± 0.0055 (AB) 0.0024 ± 0.0014 (B) 0.0089 ± 0.0040 (A) 0.00256 ± 0.00058 (B) 0.0261 ± 0.0016 (B) 0.0211 ± 0.0013 (C) 0.02933 ± 0.00088 (A) 0.0244 ± 0.0012 (B) 177.1 ± 5.5 (B) 170.8 ± 9.3 (B) 215.2 ± 7.4 (A) 146.8 ± 4.23 (C) 182.7 ± 7.8 (B) 88.1 ± 4.9 (B) 87.6 ± 5.5 (B) 111.5 ± 1.7 (A) 70.8 ± 3.0 (C) 87.0 ± 3.3 (B) 65 × 10 ± 44 × 10 (A) 79 × 10 ± 31 × 10 (A) 145 ± 28 (B) 67 × 10 ± 19 × 10 (A) 87 × 10 ± 56 × 10 (A) 77 ± 20 78 ± 15 73 ± 20 68 ± 15 83.9 ± 9.7 Porosity (%) Maximal individual pore volume (mm3) 0.0035 ± 0.0014 (B) Pore structure thickness (mm) 0.0249 ± 0.0010 (B) Pore specific surface area (mm-1) Pore fragmentation (mm-1) Pore connectivity Peel thickness (µm) 513 514 Table 3. At harvest ‘Hayward’ kiwifruit quality difference as a result of crop load and girdling. Values represent mean and standard deviation (in 515 brackets). Orchard Treatment Mass (g) DM (%) SSC (°Brix) Firmness (N) Porosity (%) 43 t/ha, with girdling 112.8 (11.9) 18.56 (0.01) 7.8 (0.9) 74.1 (6.7) 3.52 (0.36) 43 t/ha, without girdling 104.3 (11.5) 18.84 (0.01) 7.4 (0.8) 81.0 (6.6) 3.97 (0.52) 36 t/ha, with girdling 119.4 (13.3) 19.14 (0.01) 8.4 (0.7) 74.8 (6.5) 3.82 (0.64) 36 t/ha, without girdling 111.9 (11.5) 18.84 (0.01) 7.9 (0.9) 78.2 (6.8) 4.45 (1.47) 90 90 90 90 5 n 516 517 518 519 520 Table 4. Significance table showing the p-value for crop load and girdling effects on at-harvest ‘Hayward’ kiwifruit attributes. 521 p - value Orchard treatment Mass DM SSC Firmness Porosity (g) (%) (°Brix) (N) (%) Crop Loading < 0.001 0.002 < 0.001 0.138 0.323 Girdling < 0.001 0.931 < 0.001 < 0.001 0.180 0.680 0.002 0.791 0.014 0.828 Crop Loading * Girdling 28 522 9 Figure captions 523 Fig. 1. Image processing steps of X-ray micro-CT data of kiwifruit tissue: A. grey scale 524 reconstructed X-ray slice (4.4 µm pixel size, scale bar: 1 mm). B. Segmentation of pores by 525 manual threshold. C. Labeling of individual pores. D. 3-D rendering of pores. 526 527 Fig. 2. Typical X-ray microtomograph of commercial kiwifruit cultivars at eating ripeness 528 (A, 4.87 µm pixel size, scale bar: 1 mm), 3-D rendering of the pore structure (B, 400 × 400 × 529 460 voxel volume), and pore network model (C, 100 × 100 × 100 voxel volume) showing the 530 essential geometry and branching, as well as local thickness (in color bar) of pores. 531 532 Fig. 3. Pore size distribution expressed as volume fraction of individual pores as a function of 533 pore equivalent diameter. Sample volumes analysed (460 × 400 × 400 voxels, 4.87 µm pixel 534 resolution) were extracted in the region immediately adjacent to the skin of commercial 535 kiwifruit cultivars. Values were averaged from 5 fruits per cultivar. 536 537 Fig. 4. Druses within outer kiwifruit tissues visualized by micro X-ray CT. Maximum 538 intensity projection image presents the surface of a ‘G3’ kiwifruit sample showing widely- 539 scattered druses (A, bounding box: 2 mm × 2mm × 2.2 mm). Tomographic CT image shows 540 a transverse reconstructed image of ‘G3’ kiwifruit tissue with druses (B, 2 mm × 2 mm). 541 Surface rendering of typical druses from different kiwifruit cultivars (C, scale: 100 µm). 542 543 Fig. 5. Druse count per volume for each cultivar (A) and its morphometric characteristics 544 showing the mean bundle volume (B), length (C), and sphericity (D) obtained from X-ray CT 545 images (4.87 µm pixel resolution, 460 × 400 × 400 voxel volume). Values were averaged 546 from 5 fruits from each cultivar. Error bar represent 95 % confidence of the mean. 547 30 548 10 Figures A B C D 549 550 Fig. 1. Image processing steps of X-ray micro-CT data of kiwifruit tissue: A. grey scale 551 reconstructed X-ray slice (4.4 µm pixel size, scale bar: 1 mm). B. Segmentation of pores by 552 manual threshold. C. Labeling of individual pores. D. 3-D rendering of pores. 553 Cultivar A B C Hayward Hort16A G3 G9 G14 554 Fig. 2. Typical X-ray microtomograph of commercial kiwifruit cultivars at eating ripeness 555 (A, 4.87 µm pixel size, scale bar: 1 mm), 3-D rendering of the pore structure (B, 400 × 400 × 556 460 voxel volume), and pore network model (C, 100 × 100 × 100 voxel volume) showing the 557 essential geometry and branching, as well as local thickness (in color bar) of pores. 32 558 25 G9 Volume fraction (%) 20 Hort16A G3 15 G14 Hayward 10 5 0 0 559 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 Pore equivalent diameter (µm) 560 Fig. 3. Pore size distribution expressed as volume fraction of individual pores as a function of 561 pore equivalent diameter. Sample volumes analysed (460 × 400 × 400 voxels, 4.87 µm pixel 562 resolution) were extracted in the region immediately adjacent to the skin of commercial 563 kiwifruit cultivars. Values were averaged from 5 fruits per cultivar. 564 565 566 A (C) Hayward Hort16A B G3 G9 G14 567 568 Fig. 4. Druses within outer kiwifruit tissues visualized by micro X-ray CT. Maximum 569 intensity projection image presents the surface of a ‘G3’ kiwifruit sample showing widely- 570 scattered druses (A, bounding box: 2 mm × 2mm × 2.2 mm). Tomographic CT image shows 571 a transverse reconstructed image of ‘G3’ kiwifruit tissue with druses (B, 2 mm × 2 mm). 572 Surface rendering of typical druses from different kiwifruit cultivars (C, scale: 100 µm). 573 34 7 B 250000 5 Volume (µm3) Druse count (mm-3) 6 300000 A 4 3 2 200000 150000 100000 1 50000 0 180 0 0.95 160 C D 0.9 Sphericity Length (µm) 140 120 100 0.85 80 60 40 0.8 G9 574 Hort16A G3 G14 Kiwifruit cultivar Hayward G9 Hort16A G3 G14 Hayward Kiwifruit cultivar 575 Fig. 5. Druse count per volume for each cultivar (A) and its morphometric characteristics 576 showing the mean bundle volume (B), length (C), and sphericity (D) obtained from X-ray CT 577 images (4.87 µm pixel resolution, 460 × 400 × 400 voxel volume). Values were averaged 578 from 5 fruits from each cultivar. Error bar represent 95 % confidence of the mean.