Annals of Botany 81 : 697–703, 1998

Pollen Tube Distribution in the Kiwifruit (Actinidia deliciosa A. Chev. C. F. Liang)

Pistil in Relation to its Reproductive Process

D. HOWPAGE*, V. VITHANAGE† and R. SPOONER-HART*

*School of Horticulture, Uni�ersity of Western Sydney (Hawkesbury), Bourke Street, Richmond, NSW 2753,

Australia and †CSIRO Plant Industry, Horticulture Research Unit, 306 Carmody Road, St Lucia,

QLD 4067, Australia

Received: 7 October 1997 Returned for revision: 14 November 1997 Accepted: 30 January 1998

High resolution light and fluorescence microscopy were used to investigate the structural and cytochemical featuresof the kiwifruit (Actinidia deliciosa) pistil and to follow the pollen tube pathway after pollination. The multicarpellaryovary is syncarpous only at the ovary level thus leaving 30–40 free styles on top. The fusion of the longitudinally-folded carpels to form the syncarpous ovary forms a central parenchymatous axis, the columella, from which theovules radiate outwards into the ovary cavity. A prominent cup shaped depression on the columella at the top endof the ovary, termed the pollen tube distributor cup (PTDC) was detected. Pollen tubes from the stigma travel throughthe transmitting tract and enter the PTDC from where they are distributed towards the ovary. Even when pollinationis restricted to two stigmas, the PTDC seems to ensure that the pollen tubes are evenly distributed around the ovaryresulting in an even distribution of seeds. This suggested role of the PTDC which could compensate for over andunder pollination of individual stigmatic arms is another adaptive feature which plays a crucial role in thereproductive process of kiwifruit. The significance of this structure for pollination by insects is discussed.

# 1998 Annals of Botany Company

Key words : Actinidia deliciosa, kiwifruit, pollination, floral anatomy, reproductive success, pollen tube distribution.

INTRODUCTION

Kiwifruit, Actinidia deliciosa (A. Chev.) C. F. Liang andA. R. Ferguson is a dioecious fruit species native to China(Ferguson, 1983, 1984) and grown inmany parts of theworldfor its edible fruit.

In kiwifruit, staminate (male) vines produce male flowersthat contain viable pollen, but they have no functionalovaries. Pistillate (female) vines produce female flowers with

b1

b2

a2a1

st

F. 1. A female (pistillate) kiwifruit flower. Sections were taken alongthe symmetrical line (a

"a#) of the flower where styles (st) are arranged

linearly, and at right angles (b"b#) to a

"a#. Bar¯ 1 cm.

a functional ovary and non-viable pollen (Schmid, 1978).Pistillate vines bear fruit. In pistillate flowers, the ovary isformed by the fusion of 26–41 carpels with each housing upto 40 ovules with an estimated total of 1400–1500 ovules perpistil (Hopping, 1976; Gue!de! s and Schmid, 1978). Thus,there is a need to transfer an exceptionally large amount ofpollen to fertilize all ovules. Fertilization of a largeproportion of ovules results in bigger fruit (Hopping, 1976),and fruit size and seed numbers are positively correlated(Pyke and Alspach, 1986; Testolin, Vizzotto and Costa,1991).

Honey bees (Apis mellifera L.) are regarded as essentialand true pollinators of kiwifruit (Palmer-Jones and Clinch,1974; Howpage, Spooner-Hart and Vithanage, 1996;Vaissie' re et al., 1996). The wet stigma (Harvey et al., 1987)and the short pollination period measured in terms ofstigma receptivity lasts for 4 d (Gonza! lez, Coque andHerrero, 1995a, b) and is considered an adaptation forinsect pollination. However some floral characteristics ofkiwifruit, such as the pendulous nature of flowers, absenceof pollenkit, high ovule:pollen ratio and large multi-branched stigmatic surfaces suggested that wind may be anequally important agent of pollen transfer as honey bees(Craig and Stewart, 1988; Costa, Testolin and Vizzotto,1993).

The pistil is well adapted to nourish a large number ofpollen tubes as demonstrated by the copious secretionsfound in the pollen tube pathway, including the placentalsurface (Gonza! lez, Coque and Herrero, 1996), an attributelinked to its high reproductive success. In this paper, we

0305-7364}98}060697­07 $25.00}0 bo980615 # 1998 Annals of Botany Company

698 Howpage et al.—Pollen Tube Distribution in Kiwifruit

trace the path of pollen tube growth and report somestructural features of the pistil which play an additional,crucial role in the reproductive process of kiwifruit ; theirsignificance in crop production is discussed.

MATERIALS AND METHODS

Plant material

Flowers used in this study were from a 5 hectare commercialorchard in Agnes Banks (33° 37« S, 150° 41« E) NSW,Australia. The vines were 8-year-old ‘Hayward’ cultivarplanted in 4¬5 m spacing. Male vines (four–five unknowncultivars) were planted in every third position of every thirdrow. Flowering commenced on 1 Nov. 1996, and 40 honeybee colonies of Apis mellifera ligustica were introduced tothe orchard on 3 Nov. 1996 at approx. 5% flowering level(female flowers), which is normal management practice.

Controlled pollination

Over 100 flower buds from a single vine were bagged1–2 d before anthesis, using brown paper bags. Ten flowerswere subjected to each of the six pollination treatmentsdescribed below. All pollinations were carried out with afine camel hair paint brush using freshly collected pollenfrom a single male vine. The application was limited to asingle dab of the paint brush. The pollination treatmentswere: (a) hand pollinated on two adjacent stigmas leavingthe rest of the stigmas unpollinated; (b) hand pollinated asabove but on two stigmas at opposite ends of the flower; therest of the stigmas were left unpollinated; (c) hand pollinatedon two stigmas at opposite ends; unpollinated styles wereremoved from the base of the ovary using a pair of sharpentomological scissors ; (d) hand pollinated on all stigmas;(e) open pollinated control ; (f) unpollinated control. Allpollinations were carried out on the day of bagging atanthesis or 1 d after anthesis. Flowers were re-baggedimmediately after hand pollination and five flowers fromeach treatment were sampled 3 d after pollination (3 d afteranthesis for open pollinated and unpollinated controls) forprocessing and the rest were re-bagged. The bags wereremoved after 15 d to assess fruit set. Fruit was harvested on3 Apr. 1997 and fruit weight and seed distribution in crosssection were assessed. Photographs of fruit cross sectionstaken at three levels were divided into four quarters and theseeds in each quarter were counted to estimate thedistribution of seeds within the fruits produced by eachpollination treatment. These were statistically comparedpairwise using Student’s t-test.

In addition to the above flowers, ten more open pollinatedflowers from ten randomly selected vines were sampled 3 dafter anthesis (3 DAA) for processing to estimate pollengermination on stigmas under field conditions.

Tissue preparation

Specimens were fixed in 3% glutaraldehyde in 0±1

phosphate buffer (pH 7±5) for 12 h at room temperature, orin acetic :alcohol (acetic acid:ethanol, 1 :3). After washing

the pistils in 0±1 phosphate buffer (pH 7±5), they weredehydrated in an acetone series and embedded inHistoresin2(Reichert-Jung, Heidelberg).

Flowers used for sectioning were cut into convenient sizesfor ease of handling during fixation and embedding. Initially,all anthers, petals and sepals were removed from flowersand styles were cut from the base. Longitudinal sections ofthe ovary were made in two planes at right angles to eachother as shown in Fig. 1: (a) symmetrically along the midline of the ovary where all styles were arranged linearly (Fig.1, a

"a#) ; (b) at right angles to the above (Fig. 1, b

"b#). Serial

sections, 2–3 µm thick, were taken throughout each ovaryand style and stained with periodic acid–Schiff’s reagent(PAS) and toluidine blue O (TBO) (O’Brian and McCully,1981) and observed using light microscopy.

Pistils used for fluorescence microscopy were cut intoapprox. 1±0 mm thick sections along the same planes a

"a#

and b"b#(Fig. 1) prior to fixing in acetic :alcohol (1:3 v}v).

The fixed material was then softened in 10% sodiumsulphite for 40–60 min in a water bath at 60 °C, stained indecolorized aniline blue (Dumas and Knox, 1983), squashedgently on a glass slide and observed using fluorescencemicroscopy (Olympus BX 60).

RESULTS

Anatomical features of the pistil

The free styles in female kiwifruit flowers are arranged intwo rows on the ovary (Figs 1, 3A and B). The papillate

F. 2. Sections of the style 3 d after anthesis stained by PAS andcounterstained with toluidine blue, showing papillae (p), and trans-mitting tissues (t). Bars¯ 150 µm. A, Transverse section of the style atthe distal end; B, transverse section of the style at the base, showing the

closed transmitting channel.

Howpage et al.—Pollen Tube Distribution in Kiwifruit 699

F. 3. Light micrographs showing sections of the kiwifruit pistil at the ovary-stylar junction, stained with PAS. Transverse sections taken at thebase of the style and moving progressively towards the ovary (A–E). Longitudinal sections taken along two planes at right angles to each other(F–G). The cup (c) is described in the text as the pollen tube distributor cup (PTDC). A, Transverse section of kiwifruit pistil at the ovary–stylarjunction where the base of the style (st) is closed and heavily stained with PAS. Bar¯ 350 µm; B, transverse section of kiwifruit pistil taken fromjust below the ovary–stylar junction where the base of the style (st) opens into the free space. Bar¯ 350 µm; C, transverse section of kiwifruitpistil taken from above the upper carpellary level showing the widening free space to form the cup (c) and carpels (ca). Bar¯ 350 µm. s,Secretions. D, Transverse section of kiwifruit pistil taken at the cup and carpellary level, showing the cup (c) with secretions (s) and ovules (ov).Bar¯ 350 µm; E, transverse section of kiwifruit pistil taken from a lower carpellary level ; note the reduced cup size (c) and ovules (ov).Bar¯ 350 µm. F, Longitudinal section taken at the axis, a

"a#of the pistil (Fig. 1), showing cup (c) and its association with a carpel. Bar¯ 350 µm.

S, Secretions. G, Longitudinal section taken at the axis, b"b#of the pistil (Fig. 1), showing the cup (c) with secretions (s) and its association with

half of the styles. Bar¯ 350 µm.

700 Howpage et al.—Pollen Tube Distribution in Kiwifruit

stigmas are seen at the free distal ends of the style. Furthertowards the base of the styles, the carpels are longitudinallyfused to enclose the transmitting tract for pollen tubes. Thisis seen in the transverse sections at the distal and theproximal ends of the style (Fig. 2A and B).

The multicarpellary ovary in kiwifruit is syncarpous onlyat the ovarian level, hence the free styles at the distal end.Each carpel bears two rows of ten–20 anatropous ovulesalong either side of the carpellary sutures. The fusion of thelongitudinally-folded carpel bases to form the multi-carpellary ovary has left a central column of paren-chymatous tissue, termed the ‘columella ’ (Schmid, 1978). Inthe ovary, the ovules radiate outwards from the columellainto the carpellary cavity of the ovary. At the top of thecolumella, just below the free stylar zone, there is aprominent cup shaped depression which forms a cavity (Fig.3A, F and G).

The transmitting tract, which is enclosed by carpellarytissue near the base of the styles (Fig. 3B–D), opens into thiscavity which we describe as the pollen tube distributor cup(PTDC) (Figs 3C, D, F, G, 4A, C, D, and 5).

Pollen tube pathway

The pollen tube pathway is clearly supplied with copiousamounts of secretions containing pectinaceous poly-saccharides as revealed by PAS and toluidine blue staining.Upon pollination, the pollen tubes grow through thetransmitting tract as seen in TS of styles (Figs 3A, B, C and4B). In the upper region of the ovary, the individualtransmitting tracts open into one large space which iscontinuous with the ovary cavity (Fig. 3C, F and G). Oncethe tubes leave the individual transmitting tracts, they thenenter this PTDC (Fig. 4A, C and D) before growing overthe rim to enter the carpellary cavity, towards the ovules.This is seen clearly in hand pollination experiments withlimited amounts of pollen (Fig. 5). In normal open pollinatedsituations, the entire PTDC is filled with growing pollentubes (Fig. 4A, C and D).

No matter whether adjacent or opposite stigmas werepollinated, pollen tubes reaching the ovary remained evenlydistributed around the ovary. In the treatment in whichunpollinated stigmas were removed (treatment C), pollentubes entered the PTDC (Fig. 5) and from there they weredistributed evenly around the ovary and entered individualovules effecting fertilization (Fig. 4D and E). The PTDC isalso heavily lined with secretory material containingpectinaceous polysaccharides as revealed by PAS and TBOstaining (Fig. 3C, D and G). This suggested functionalfeature of the PTDC is further confirmed by the seeddistribution of fruits produced in our hand pollinationexperiments. Student’s t-test applied to seed data fromdifferent fruit quarters showed no significant departure(P" 0±31) from those of any other quarter (Table 3).

Pollen germination and fruit}seed set

Squashed preparations of open pollinated styles collected3 DAA showed that the deposition and subsequent ger-

mination of pollen between stigmatic surfaces within thesame flower was uneven (Table 1). Some stigmas had highnumbers of pollen grains germinating—up to 300—whileothers had none.

All pollination treatments, including open pollination,produced 100% fruit set except in the unpollinated controlswhere there was no fruit set. Although seed distribution wasunaffected by a low level of pollination, fruit size and weightwas affected (Table 2).

DISCUSSION

Our results show the existence of a special cup-shapedstructure on the central axis of the ovary, the pollen tubedistributor cup, which appears to be involved in distributingpollen tubes evenly around the ovary, even when pollinationis restricted to two stigmas. The significance of this cup-shaped structure and its suggested function seems to be inthe even distribution of seeds around the fruit core. Whilstobservations of in �i�o pollen tube growth and seeddistribution data seem to support this hypothesis, otherfactors such as ovule signals and nutrient gradients may alsohave an influence on the random seed distribution inkiwifruit.

Pollination is an important pre-requisite for fertilizationand is also influential in maintaining fruit shape anduniformity via its effect on seed production not only inkiwifruit fruit (Lawes, Wooley and Lai, 1990) but also inother fruits such as feijoa (Patterson, 1990), nashi fruit(Rohitha and Klinac, 1990) and strawberry (Svensson,1991). Whilst the biological significance of an evenly shapedfruit is speculation at this stage, it has commercialsignificance in that it guarantees a quality product in termsof shape, even when pollination conditions are limiting.This contrasts with observations made in some multi-seededfruit, e.g. Annona (Sanewski, 1988), where inadequatepollination results in misshapen fruit.

Honey bees play a major role in kiwifruit pollination.Under orchard conditions, the low attractiveness of theflowers to bees (Jay and Jay, 1984), low levels of pollenproduction in the afternoon (Goodwin, 1995), low levels ofbee cross over (Goodwin and Steven, 1993), less pollencarryover in bee corbiculae (Goodwin and Perry, 1992) andthe release of pollen in clumps (McKay, 1976) havecontributed to uneven and inadequate pollination ofstigmatic surfaces. Theoretically, a kiwifruit flower needs atleast 50 pollen grains per style to produce a fruit with 1400seeds, provided each stigma receives an equal amount ofpollen. However, this is not always the case. We haveobserved hundreds of pollen tubes growing through thetransmitting tract within a single style (Fig. 4B), while otherstyles had none. In such cases, seed distribution is even andthe fruit shape regular indicating that there is a mechanismto compensate for any uneven pollination by ensuring aneven distribution of seeds. This mechanism is also likely tobe advantageous to the developing seed ensuring maximumspacing is available to each. The kiwifruit system contrastswith that described for maize by Heslop-Harrison, Heslop-Harrison and Reger (1985) in an elegant series of exper-iments in which pollination of a sector of the maize silk

Howpage et al.—Pollen Tube Distribution in Kiwifruit 701

F. 4. Fluorescence micrographs of open pollinated kiwifruit pistils showing the pollen tube pathway (stained with decolorized aniline blue). A,LS view at the a

"a#(Fig. 1) axis of the pistil ; pollen tubes compacted in the cup (c) ; the cup is described in the text as the pollen tube distributor

cup (PTDC). Bar¯ 150 µm; B, hundreds of pollen tubes (pt) passing through the transmitting tissue of the style (st). Bar¯ 150 µm; C,longitudinal section along the symmetrical axis b

"b#(Fig. 1), showing pollen tubes compacted within the cup (c). Bar¯ 150 µm; D, pollen tubes

that come out of the cup have reached the ovules. Bar¯ 150 µm; E, a pollen tube (pt) entering an ovule (ov) through the micropyle (m); notethe funicle (f). Bar¯ 70µm.

702 Howpage et al.—Pollen Tube Distribution in Kiwifruit

F. 5. Fluorescence micrograph of the ovary–stylar region of a kiwifruit pistil where only two stigmas at opposite ends were pollinated; the fewpollen tubes have reached the cup (c) and from there are distributed towards the ovary; some pollen tubes (pt) have reached the ovules (ov) ; the

cup (c) is described in the text as the pollen tube distributor cup (PTDC). Bar¯ 150µm.

T 1. Number of pollen grains germinating on stigmas ofopen pollinated kiwifruit flowers

Flower Mean³s.e.no. (n¯ 10) Minimum Maximum

1 28±1³9±9 0 902 42±7³21±5 3 2303 4±7³0±8 2 114 47±7³6±7 9 815 66±6³27±1 2 2506 12±8³3±7 2 387 88±1³29±2 1 3008 35±4³17±7 0 1809 10±2³5±3 0 55

10 15±9³6±4 0 64

T 2. Mean fruit weight (g) resulting from differentpollination treatments

Pollination Mean fruittreatment weight³s.e.

1. Adjacent stigmas 51±2³11±72. Opposite stigmas 53±25³10±03. Opposite stigmas 43±2³8±0

others removed4. Open pollination 91±6³4±7

Treatment 1, two adjacent stigmas were pollinated while the rest ofthe stigmas received no pollen; treatment 2, two stigmas at the oppositeends of the flower were pollinated while the rest of the stigmas receivedno pollen; treatment 3, as for treatment 2 except that the unpollinatedstyles were cut off; treatment 4, all stigmas were pollinated; anunpollinated control (data not shown) had no fruits.

T 3. Mean seed number counted in each quarter of fruitTS taken from the top, middle and bottom of each fruit

obtained from the different pollination treatments

Mean seed number counted in eachquarter of the fruit from three layers

of the cross section

Pollination 1st 2nd 3rd 4thtreatment quarter quarter quarter quarter

Adjacent stigma 6±6 4±3 5±3 7±0Opposite stigma 4±6 3±0 2±6 3±6Pruned stigma 4±6 3±6 2±3 3±3Open pollination 6±6 6±3 5±0 4±6

Pairwise comparison of seeds in each quarter for each treatmentusing Student’s t-test did not show any significant deviation from thoseof any other quarter (P" 0±31).

(stigma) led to fertilization of only that sector of the cob. Inkiwifruit, this may have a special adaptive significance notonly towards pollinating agents but also towards the seeddispersing agents in its native habitat, China.

This special structure, the PTDC, therefore, adds to theother floral features such as abundant flowering, multi-branched stigmatic surfaces and copious secretions in theovary system (Gonza! lez et al., 1996) that contribute towardsthe reproductive success of kiwifruit. Our study alsoconfirms the presence of copious, pectinaceous secretions inthe ovary system as localized by PAS and TBO staining inthe stylar transmitting tract, the PTDC and the base of theovules. Such secretions are found along the pollen tube

Howpage et al.—Pollen Tube Distribution in Kiwifruit 703

pathway of most angiosperms (Knox, 1984) in varyingamounts, as revealed by staining.

Low levels of pollination followed by fertilization usuallyproduce smaller fruit, a result confirmed by our study aswell as those of others (McKay, 1976; Pyke and Alspach,1986). In our experiments hand pollinations were carriedout with only a single dab of the paint brush to ensureuniform transfer of a small quantity of pollen. However,this did not restrict the seed development to only one sectoras observed in other fruit, thus confirming our earlierhypothesis. Whilst wind pollination would allow an evendistribution of pollen on the stigmas, which in turn wouldlead to an even fruit, the mechanism in kiwifruit seems toachieve the same result even if an insect pollinates only onepart of the style. This can be considered a true adaptation ofthis flower for pollination by insects such as honey bees.

ACKNOWLEDGEMENTS

We thank Ms Arkey James for providing access to the studyproperty. We also thank the University of Western Sydney(Hawkesbury) for funding this research and Mr GregTurnbull for taking photographs of plates. We also thankMr V. Premajayanthe for statistical advice and Ms BarbaraMay for drawing Fig. 1. D. Howpage was funded by aHawkesbury Postgraduate Award.

LITERATURE CITED

Costa G, Testolin R, Vizzotto G. 1993. Kiwifruit pollination: anunbiased estimate of wind and bee contribution. New ZealandJournal of Crop and Horticultural Science 21 : 189–195.

Craig JL, Stewart AM. 1988. A review of kiwifruit pollination: whereto next? New Zealand Journal of Experimental Agriculture 16 :385–399.

Dumas C, Knox RB. 1983. Callose and determination of pistil viabilityand incompatibility. Theoretical and Applied Genetics 67 : 1–10.

Ferguson AR. 1983. E. H. Wilson. Yichang, and the kiwifruit. Arnoldia43 : 24–35.

Ferguson AR. 1984. Kiwifruit : a botanical review. Horticultural Re�iews6 : 1–64.

Gonza! lez MV, Coque M, Herrero M. 1995a. Stigmatic receptivity limitsthe effective pollination period in kiwifruit. Journal of AmericanSociety of Horticultural Science 120 : 199–202.

Gonza! lez MV, Coque M, Herrero M. 1995b. Papillar integrity as anindicator of stigmatic receptivity in kiwifruit (Actinidia deliciosa).Journal of Experimental Botany 46 : 263–269.

Gonza! lez MV, Coque M, Herrero M. 1996. Pollen pistil interaction inkiwifruit (Actinidia deliciosa ; Actinidiaceae). American Journal ofBotany 83 : 148–154.

Goodwin RM. 1995. Afternoon decline in kiwifruit pollen collection.New Zealand Journal of Crop and Horticultural Science 23 :163–171.

Goodwin RM, Perry JH. 1992. Use of pollen traps to investigate theforaging behaviour of honey bee colonies in kiwifruit pollination.New Zealand Journal of Crop and Horticultural Science 20 : 23–26.

Goodwin RM, Steven D. 1993. Behaviour of honey bees visitingkiwifruit flowers. New Zealand Journal of Crop and HorticulturalScience 21 : 17–24.

Gue! de! s M, Schmid R. 1978. The peltate (Ascidate) carpel theory andcarpel peltation in Actinidia chinensis (Actinidiaceae). Flora (Jena)Abt. B 167 : 525–543.

Harvey CF, Fraser LG, Pavis SE, Considine JA. 1987. Floral biology oftwo species of Actinidia (Actinidiaceae). 1. The stigma. Pollinationand fertilization. Botanical Gazette 148 : 426–432.

Heslop-Harrison Y, Heslop-Harrison J, Reger BJ. 1985. The pollen–stigma interaction in the grasses. 7. Pollen-tube guidance and theregulation of tube numbers in Zea mays L. Acta BotanicaNeerlandica 34 : 193–211.

Hopping ME. 1976. Effect of exogenous auxins, gibberellins, andcytokinins on fruit development in Chinese gooseberry (Actinidiachinensis Plunch.). New Zealand Journal of Botany 14 : 69–75.

Howpage D, Spooner-Hart R, Vithanage V. 1996. A study of kiwifruit,Actinidia deliciosa cv. Hayward pollination in Australia. In: PlantReproduction 96, Abstracts, 14th International Congress of SexualPlant Reproduction. Melbourne: University of Melbourne Press.

Jay D, Jay C. 1984. Observations of honey bees on Chinese gooseberries(kiwifruit) in New Zealand. Bee World 65 : 155–166.

Knox RB. 1984. Pollen–pistil interactions. In: Linskens HF, Heslop-Harrison J, eds. Cellular interactions. Encyclopedia of PlantPhysiology 17 : 508–608.

Lawes GS, Wooley DJ, Lai R. 1990. Seeds and other factors affectingfruit size in kiwifruit. Acta Horticulturae 282 : 257–264.

McKay SA. 1976. Pollination and other factors affecting fruit set and sizeof kiwifruits. MSc Thesis, University of California, Davis.

O’Brian TP, McCully ME. 1981. The study of plant structure: principlesand selected methods. Victoria, Australia : Termarcarphi Pty Ltd.

Palmer-Jones T, Clinch PG. 1974. Observation on the pollination ofChinese gooseberries variety ‘‘Hayward’’. New Zealand Journal ofExperimental Agriculture 2 : 455–458.

Patterson KJ. 1990. Effects of pollination on fruit set, size and qualityof Feijoa (Acca sellowiana (Berg) Burret). New Zealand Journal ofCrop and Horticultural Science 18 : 127–131.

Pyke NB, Alspach PA. 1986. Inter-relationships of fruit weight, seednumber and seed weight in kiwifruit. New Zealand Journal ofAgricultural Science 20 : 153–156.

Rohitha BH, Klinac DJ. 1990. Relationship between seed set and fruitweight and shape of nashi (Pyrus serotina Rehder var. cultaRehder). New Zealand Journal of Crop and Horticultural Science18 : 133–136.

Sanewski GM. 1988. Growing custard apples. Brisbane: QueenslandGovernment.

Schmid R. 1978. Reproductive anatomy of Actinidia chinensis (Actin-idiaceae). Botanischer JahrbuX cher fuX r Systematik 100 : 149–195.

Svensson B. 1991. The importance of honeybee-pollination for thequality and quantity of strawberries (Fragaria ananassa) in centralSweden. Proceedings of the Sixth International Symposium onPollination, Tilburg, Netherlands, 27–31 Aug. 1990, pp. 260–264.

Testolin R, Vizzotto G, Costa G. 1991. Kiwifruit pollination by windand insects in Italy. New Zealand Journal of Crop and HorticulturalScience 19 : 381–384.

Vaissie' re BE, Rodet G, Cousin M, Botella L, Torre! Grossa JP. 1996.

Pollination effectiveness of honey bees (Hymenoptera: Apidae) ina kiwifruit orchard. Journal of Economic Entomology 89 : 453–461.