THE UNIVERSITY OF CALGARY

klaternal Investment. Pollination EtEciency and Pol1en:Ovuie Ratios in Alberta Orchids

blagdalena J. Lukasiewicz

A THESIS SLBkIITTED TO TKE F-ACLrLTYOF GRADUATE STUDIES IN PARTI-AL FLLFLLLMENT OF THE REQLiTREkCEhTS FOR THE DEGREE OF IMASTER OF SCIENCE

DEP.ARTkENT OF BIOLOGICAL SCIENCES CALGARY, ALBERTA

AUGUST, 1999

0Magdalena J. Lukasiewicz I999

1*1

National Library of Canada

Biblioth&que nationale du Canada

Acquisitions and Bibliographic Services

Acquisitions et services bibliographiques

395 Wellington Street OttawaON KlAON4 Canada

395. rue Wellington OctawaON KlAON4 Canada

Your rYI V U m M -

our

rn mtre m

w m

The author has granted a nonexclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis in microform, paper or electronic formats.

L'auteur a accorde une licence non exclusive pennettant a la Bibliotheque nationale du Canada de reproduire, prster, distribuer ou vendre des copies de cette these sous la forme de microfiche/film, de reproduction sur papier ou sur format electronique.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts from it may be printed or othenvise reproduced without the author's permission.

L'auteur conserve la propriete du droit d7auteurqui protege cette these. Ni la these ni des extraits substantiels de celle-ci ne doivent Stre imprimes ou autrement reproduits sans son autorisation.

This study investiyated patterns of reproductive resource allocation between female and

male hnction and within female function for 17 temperate orchid species. Resource allocation between the competing sex roles and within a sex hnction was considered. I

-

investisated opportunities for adjustment of maternal resource investment durin, one reproductive season- Because orchid obuies are immature at anthesis and hrther

de\-elopment is triggered by pollination, orchids may demonstrate unique opportunities for adjustment of ovule number to pollen receipt. However. like other angiosperms orchids did not adjust ovule number to pollen load. Instead. seed number. but not size. responded to pollen quality. I also considered the relation between resource allocation. P . 0 ratio and pollination efficiency. Compared to species with granular pollen, a very hizh proportion ( 10409'0) of pollen removed from orchid flowers reaches stigmas

ipollinarion efficiency). Correspondingly. orchids have considerably lower P:Othan angiosperms with granular poIIen. Within orchids and among angiosperms P:Oratio x-aries negatively with pollination efficiency. These results are discussed in light of the eft'ects of increased pollination efficiency on local mate competition and resource allocation.

1 would like to thank. above all. Lawrence Harder my supervisor for his continued

patience. encouragement. invaluable theoretical assistance, and statistical expertise.

Lawrence was always approachable, supportive and esact in his expectations, suggestions and advice especially throughout the writing of my thesis. The other members of my

committee, hlary Reid and Ed Yeuny, contributed valuable feedback during the development of my study. In particular. Ed Yeung provided much appreciated equipment and technical advice. I could not have completed my data collection with out considerable help from

Taline Sarkissian. whose hard work and easygoing nature allotved tbr examination of so many species. I must also acknowledpe Eda Czamecki, for her company and infallible

l my other lab mates, Jen humour during our orchid hunting excursions. 1 am ~ r a t e h to L\fVaIker-Larson and Crispin Jordan. tbr their companionship and enthusiasm for discussion.

I thank my parents. Stanislaw and Krystyna Lukasiewicz. for encouragement and support throughout my academic career. hly mother's continued interest in my progress and my father's invariable faith sustained my resolve. I also appreciate discussions (from an engineer's perspective) with Krzysztof Paika. his particular sense of humour and his tremendous enthusiasm. which greatly contributed to t h e successful completion of this thesis.

In memory of Marianna Kolbow

TABLE OF CONTENTS

PAGE .. X P P R 0 V . U PAGE ~ . ~ . . ~ ~ - . . . . . . . . . . . . . ~ - . - - - ~ . . - ~ ~ . ~ . ~ ~ ~ - . - . ~ . . . .11- - . . . . . . . . . . . . . . .

.ABST R.4CT

...........................

. . . .................................................

..

111

.A CKYOCVLEDGEMENTS ................................................................

iv

DEDLC.ATl0N

v

..............................................................................

T.U3LE Of CONTENTS

T.M3I-E OFT.Al3LES

TABLE OF FIGURES

1

...................................................................

vi

........................................... . . . .

ix

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

si

...................

REPRODLCTION . RESOLrRCE .ULOCATION .Xh i i ORCHIDS I 1 Introduction

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........

1.2 Adjustment of XIaternaI Investment

2

.........................

...........................................

1.3 Resource .Allocation Between the Sex Functions

I IObjectives

............

......................................

..............................

. . . ..............................

1 1 3

7 11

REPRODUCTIVE FE-ATURES OF ORCHIDS .41\iD A DESCRIPTIOX OF

STLDY SPECIES

...................................... . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1 Unique Reproductive Traits of Orchids

13

........................................

13

2.1.1 Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

2.1.2 PolIinia .......................................................................

I4

2.1.3 Ovules ........................................................................

20

2.1 .-! Seeds .........................................................................

20

3.2 Investiyated AIberta Orchid Species ............................................

3

OPPORTL%-ITIES FOR ..U>JUSmCENT OF ,MATERi.NAL RESOURCE INCrESTkLENT IN THE ORCHIDACEAE

......................

................ . .

3.1 1NTRODC;:CTION .................................................................

3 . 1 . l Adjustment of Maternal Resource Investment

3.1.2 Objectives

.........................

...................................................................

3.2 METHODS ....................... . . ......... . . .................................

3.2. I Study Species and Sites

.

3 7

2 ..

Pollination Treatments

................................................... .....................................................

3.2.3 Ovule and Seed Counting ................................................. 3 2 . 4 Ocule Fate ................................................................... 3 2 . 5 Statistical Analyses

.......................................................

................ 3 .3.1 Ok-uItt Production ...................................... . . .

3.3.2 Fruit Production ............................................................. 3.3.3 Proportionai Seed Set

.....................................................

3 3 . 4 Ovule Fate ...................................

.... ............................

3 -3.5 Seed Volume ................................................................

21

4

RESOURCE ALLOCATION AS INDICATED BY THE POLLEN:OwLE ELATIO

ORCHIDS: EFFECTS OF EFFICIENT POLLINATION ...........

4.1 J3TRODUCT[ON ................................................................. 4.1. I Resource .Allocation and Pollen:Otule Ratio ...........................

. 4.1.2 Objectives *

........................ . . . .....................................

4.3 METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Pollen Production. Pollen Size and Po1Ien:Ovule Ratio ...............

4.2.2 Pollination Efficiency ...................................................... 4.2.3 Statistical .A nat yses 4.3 RESLZTS

....................

................................. . .

....................... . . . ...............................................

4.3.1 Pollen Production and Size ..................................... ...........

4.3.2 Pollination Characteristics

................................................

........................ -4.3.; Pol1en:Ocule Ratio ............................ . . .

4.4 DISCUSSLON ...................................................................... 1.4.1 Pollination Efficiency ...................................................... 4.4.2 Po1ien:OtuIe Ratio .........................................................

LITERATURE CITED APPENDIX 1

APPEhDIX 2

.....................................................................

.................................................................................

........................................... .................................. . .

132 152 154

TITLE

PAGE

The relative abundance o f pollinia types in the Orchidaceae.

...........

Ovule production in response to four pollination treatments.

............ 40

Analysis o f the effects o f pollination treatments (natural, hand crosspollinated and hand o r autonomous self-pollinated) on ot-uIe production. proponional seed set and seed volume.

......................

.Analysis of the ettPcts o f mating type and pollination treatment on o w l e production. seed proportion and seed volume in four, ene era (C)pr.iprcIirrm. PILrtorrrher-cr,Lisrrrcr. Col.c7//orhizc1) represented by

species with contrasting mating types.

.................................... ..

Analysis of the effects o f pollination treatment. genus, and species on ovule production, seed proportion. and seed volume in xenogamous species.

..........................................................................

Mean proportional fruit ser in response to three experimental pollination treatments.

........................................................

The effects of pollination treatment. senus and species on fruit set in xenogamous and autogarnous species.

.......................... . ........

The proportion o f otvles developin2 into seeds in response to fbur pollinat ion treatments. .........................

. . ...........................

Proportion o f empty ocules due to overproduction and loss o f e z b q - o / fertilization faiIure. ..........................................................

15

3.9

Relation of the volume of individual seeds to pollination treatment. ... 69

3 10

Estimates of the effects of pollination treatments. ovule production and proportional seed set on seed voiume. .................................

-4 I

Pollen production, pollen diameter. ovule production and P:O ratio in 17 orchid species. .............................................................

4.2

71

99

.Analysis of the effects of polliniurn type. mating type, genus and species on poilen production, pollen diameter, and P:O ratio for 17

orchid species. 1.3

...............................

. ...............................

The mean proportion of flowers per inflorescence which esperienced pollen removal and/or deposition and mean pollination efficiency.

4.4

102

....

109

Analysis of effects on the proportion of flowers that experienced removal and/or deposition, and pollination efficiency.

..................

1 11

TABLE OF FIGURES

TITLE Esamples of orchid poIIinaria.

................................................

Differences in mean o u l e production between species of opposing mating types tvi t h i n four genera; Corcrllorl~i=n.Lr-srern. Plarcrlr/rrra

.

.

and C ypl-ryrc/rron. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . .. . . . . . . . . . . . . . . . . . . . . . .

TIle effect of senus on mean obule production in seno,aarnous orchids.

The effects of experimental pollination treatments on mean proponional fmit set in (a) autogamous and (b) xenogamous orchids. ... The effect of mating type on mean seed proportion for (a) naturally pollinated and ( b ) self-pollinated congeneric species pairs.

...............

The effects of pollination treatment (natural. hand-cross. or hand-sel f) and genus on mean proportional seed set in senogamous species. . . . . . .

The relation of mean seed volume to (a) pollination treatment and (b) mating type among congeneric species pairs with contrasting mating types.

..............................................................................

The effects of poll inat ion treatment (natural, hand-cross. or hand-sel f) and genus on mean volume of seeds produced by senogamous species. EtTect of extreme pollination efficiency on optimal allocation to female and male hnction. .............................................................

The effects of (a) pollinium type and mating type on mean pollen production and (b) the differences in mean pollen production among genera with the same pollinium type.

.........................................

103

Differences in mean pollen diameter among pollinia types, and among

-rrenera with the same pollinium type.

............................. . . . ......

106

The effects ofpollinium type and matins type on (a) the mean proportional pollen removal and/or deposition and (b) the mean pollination efficiency.

. . ......................... . ...........................

1 12

Differences in the mean proportional pollen removal and/or deposition among genera with the same pollinium type.

...............................

1 13

The effects on mean P:O ratio of (a) pollinium type and mating type and (b) senus within pollinium type.

. . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..

I 17

The predicted and observed effscts of pollination efficiency on adjusted P:Oratio in autogarnous and senogamous species.

.............

1 20

The relation between P:O ratio and pollination efticiency for species with granular pollen. pollen a~gresation.and orchid pollen from the literature and this study.

............................... . . . .................

sii

123

I

REPRODLKTION. RESOURCE .4L,LOCATION A h D ORCHIDS

1. I

Introduction The fitness of hermaphroditic individuals is influenced by the allocation of

resources within a three-tevel hierarchy. Resource distribution to reproduction versus other bnctions initiates this structure. As fitness depends on both male and female reproductive success. allocation between the competing sex functions follows. Lastly. division and adjustment of resource investment within each sex role ultimately controls reproductive success. Resource allocation to current reproduction must balance the trade-off bemeen present reproduction and hture growth. sumival, and reproduction. Alt houyh increased reproduction during a specitic season may increase fitness. it may also be associated with substantial reproductive costs ( c g . Primack and Stacy I 998). Differential allocation bet\\-ernseasons may allow the plant to build up energy reserves needed for flowering in subsequent !-ears. In seneral. the optimal investment in reproduction balances the relati\-ecertainty of current offspring against the risk that an individual will survive to reproduce a g i n (Roff 1992).

Sex allocation involves investment in pollen production and dispersal (male tirnction) versus owle production. fertilization and seed maturation (female function) (Sutherland and Delph 1984; Wilson rt 01. i 994). Because the resources available for reproduction are finite. any increase in resource allocation to one ses role must be accompanied by decreased allocation to the other role (c.g. blazer er rrl. 1999). In

-3 -general. the optimal resolution of this tradeoff results when the marginal fitness returns through female and male hnction are equal (Charnov 1981; Lloyd 1984). .Allocation patterns within sex functions also influence a n individual plant's fitness. In general. the success of paternal allocation is difficult to examine because of the dift-cuities of following pollen dispersal (reviewed Snow and Lewis 1993) and of idcntifj-ing the fathers of seeds in natural populations. By comparison. the consequences of maternal allocation are easily studied. As a maternal parent, an individual plant has several opportunities to regulate resource expenditure on its offspring. involving both the number of matured fruit and the number of obules and seeds within each h i t (Lloyd 1930). Typically the most economic use of maternal resources involves sequential adjustment of resources invested during flowering seed development and maturation (Lloyd 1980)

The Orchidaceae provide several unique opportunities for investigating reproductive a1locat ion patterns. Several features clearly distinguish and unite this family. despite its unmatchedlspecies diversity (Dressler 1986). Orchids have evolved an elaborate pollination mechanism revolving around the hnction of coherent aggregations

of pollen (pollinia) in pollen transport between individuals (-Arditti 1993,; Dressler 1993). In addition, floral development in this group is uniquely characterized by the suspension

of ocule development until it is trigsered by pollination, so that during flowering ovaries lack mature o\ules (Withner r / cd. 1974; Nadeau er crf. 1996; Yeung and Law 1997). Furthermore. seed development in orchids is associated with limited maternal nutrient investment and the production of muItitudes of tiny seeds uniquely dependent on

mycorhizal associations for germination (.Arditti 1992: Rasmussen 1995). The combination of these features within one group of species provides an excellent opportunity to investigate resource allocation between the sex firnctions and resource investment within female hnction. 1 2

Adjustment of Maternal Resource Investment During individual reproductive seasons maternal plants have several opportunities

to adjust resource expenditure on gamete production and seed development (see Lloyd 1930: Lloyd rt a/. 1980; El Kablawy and Lovett-Doust 1996). The tirst adjustment occurs when tloral initiation and development determine the number and location of tlowers produced. In most angiosperms the ovary and ovule(s) develop before anthesis so that the female reproductive organs are hl1y mature at anthesis. Second. afier anthesis a plant may abort or develop each fruit in response to the quality and quantity of pollination. the riming of h i t initiation and the h i t ' s proximity to resources (Stephenson 195 1 ; Winsor er a/.1987; Ehrlen 199 1 : Brunet 1996). Finally. resources may be allocated variably among retained ovaries within a fruit. altering both final seed

number and size (Harper 1977). X plant may abort or invest in seeds based on genetic quality. location within the fruit and conditions during development (see Lalonde and Roitberg 1959; Mazer f 989). In general. resource r e ~ l a t i o nafter tlo~verinitiation involves downward adjustment of resource input. except perhaps during ultimate allocarion to seed size. In contrast to the standard angiosperm pattern, orchids (Orchidaceae) possess

several traits that allow unique opportunities to adjust maternal investment. In most

4

flo~veringplants ocules have matured and the egg cell is ready to be fertilized at anthesis, whereas the orchid ovary is immature and lacks mature owles at this stage (see section 2.1.3). .Although orchid species differ in the developmental stage of their ovules at anthesis. if obule decelopment has been initiated it arrests prior to meiosis after the formation of the rnegasporocyte in almost all species (Fredrikson, 1991; Nadeau et al, 1996;

see Yeung and Law 1997). Ovule development does not proceed further until

stimulated by pollination l Withner rr a/- 1974; -4rditti 1993; Zhang and O'Neill 1993; Rasmussen 1995). Poilen tubes grow down the style and then stop until the ovules develop hlly and fertilization can proceed (-4rditti 1992: Law and Yeung 1993). This delay in ovule development creates an opportunity for resource conservation or redistribution (Yeung and Law 1989; .&ditti 1992: Rasmussen 1995; Nazarav and Gerlach 1997). because the number of ovules matured could be adjusted to the number of pollen grains received by the stisma. Such a pattern may allow upward adjustment, as the resources put into otule production could be increased to match a high pollen load.

The resources saved by not producin~more ovules than can be fertilized could then be divened to other functions (Nazarov and Gerlach 1997). Whether orchids actually

implentent this conservation option is not known. Because of the potential to adjust o w l e production to pollen receipt. variation in poll inat ion could result in variable ovule production by orchids, rather than variable proportional seed set as in other angiosperms (see Burd 1994). Owle response could also retlect pollen quai ity, with self-pollination resulting in lower ovule production than cross-pollination due to maternal recognition of pollen tube quality (mate choice). If

5

ovule number is adjusted to match the number of potential fertilizations, very high

uniform proportional seed set should result irrespective of pollen quality and quantity. In addition to delayed ovule development. other features of orchid pollination may create characteristic patterns of maternal resource adjustment. In particular, orchid pot len is transponed in masses (poll inia: Dressler 1 993) (see section 2.1.2). rat her than as independent grains, so rhat a single pollinator visit delivers ample pollen to fertilize many ovules (.Arditti 1992; Rodrisuez-Robles rl a/. 1992; Dressler 1993; Proctor and Harder 1994: Wilson rr d. 1994: Xazarov and Gerlach 1997). Hence. one pollinator visit is sufficient to induce fruit maturation, resulting in production o f many seeds (Proctor and Harder 1994: Nazarov and Gerlach 1 997). Consequently. even if adjustment of ovule

number does not occur. such high pollen receipt should result in high proportional seed set in pollinated flowers. .Although orchids have high potential for prolific seed production (see section 2.1.-!I. m a n

out cross in^ (xenogamous) species. especially those that rely on deceitfbl

pollination, receive insutlkient pollinator visits to result in f u l l h i t set (Schemske 1980; hfontalc-o and Ackerman 1957; Zimmerman and .Aide 1989: Ackerman 1989; Ackerman and ;\-.Iontaivo1990:Calvo 1993). Fruit set by orchids ranges from (1%

to 8556;

hoivever. in most species pollen receipt consistently Iimits annual fruit set. as evidenced b>. the increased fruit set upon experimental hand-pollination (see Burd 1994). Because

pollination with pollinia results in high genetic similarity of orchid offspring. the maternal parent may benefit more from controlling resource investment in seeds by wl~ole-fruitabortion than by individual embryo termination (Kress 198 1 ). This

mechanism of resource conservation should occur more commonly in plants with multiple-tlowered intlorescences. than in sinsle-flowered plants. as multiple fruits provide alternate opportunities for reproductive success. Secondary to ~vhole-fruitabortion. orchids may adjust resource investment during seed production. Orchid capsules with mature seeds ofien also contain seed coats (testae) lacking a developed embryo (Catling 1952: Schmidt and -4ntlfinger 1992).The percentage of these embryo-less 'seeds' has been used to evaluate cross- or selfcompatibility in terrestrial orchids (Kallunki 198 I . Xckerrnan 1989: Schmidt and Xntlfinger 1992). tn most cases empty testae are considered as infertile seed. without explanation of their formation (Sort 1973; Ackerman 195 1; Xckerman 1989; Ackerman and llontalvo 1990; Robertson and Wyatt 1990: Tremblay 1991: Johnson and Steiner 1997. Light and XlacConnail 1998). However. these failed ocules could result tiom

several causes. including inadequate pollination. unsuccessfLl fenilization or embryo abortion. The final stase for maternal resource investment adjustment involves a sizenumber trade-otf between producing fewer larse seeds or more small seeds when

maternaI resources limit total seed mass of individual h i t s (Smith and Fretwell 1973; Harper 1977: Galen er a/. 1985: Lloyd 1957; Venable 1992). Orchid ovaries produce from hundreds to millions of ocules. which. once fertilized. form tiny (0.07- 0.1 Imm) dust-like seeds. Each seed comprises a small undifferentiated (100 cells on averase) embryo surrounded by a thin. generally transparent, seed coat (Harvais 1971; Arditti ef LI/.

1979; Arditti 1992; Dressler 1993; Rasmussen 1995). The small size of orchid seeds

7

retlrcts their reliance on an obligate symbiosis with % n l i for successful germination and early yrom-th of seedlinys (.Arditti 1992; Rasmussen 1995). This rnycotrophy, enables l i mired maternal investment in each seed. result ins in numerous tiny seeds with

negligible reserve nutrients. including reduced embryo differentiation, and reduction and eventual loss of double fertilization and endosperm formation (Benzig 198 1; Law and Yeung 1989: k d i t t i 1992; Rasmussen 1995). .As a result. seed size within orchid species should be relatively constant as the lack of substantiai nutritive resen-es within the seed limits options for increased or decreased maternal investment in each seed. Hence, pollen quality should not affect seed size within orchid species. 1.3

Resource .Allocation Berween the Sex Functions Being hermaphrodites. most angiosperms must divide limited reproductive

resources bet~veenfemale and male hnctions, so that a plant's investment in male hnction should va? inversely with investment in female function. X plant's particular division of resources between the sex roles and the timing of reproduction and sex presentation, hndamentally determine its reproductive success. The relative efectiveness of any pattern of sex allocation depends on features of a plant's reproductive environment. including mate and resource availability, pollination mechanism and frequency, and various other factors. Hence. plants with similar reproductive environments should allocate resources similarly to male and female function.

The ratio of pollen grains to ovules (P:O) produced by a flower, provides one indicator of the relative allocation of resources to male (pollen) and female (ovules)

8

functions. Angiosperms exhibit a remarkable diversity in P:O ratios (see Cruden 1977; Schoen 1977: Cruden and Jensen 1979; Cruden and Miller-Ward 198 I ; Preston 1986; t-asek and Weng 1988; Herrera 199 I ; .Armstrong 1992; Sharma er nl. 1992; Beardsell er LI/.

1993; Knudsen and Olesen 1993: Ramsey 1993; Boaz el a/. 1994; Gallardo er a/.

1994: Damgaard and Abbot 1995; Neiiand and Wilcock 1995: Jacqueman and Thompson

1996: LifBnte 1996; Nazarov and Gerlach 1997). Based on extensive surveys. Cmden ( 1977) demonstrated that

outcrossing species had higher P:O ratios (senerally

>SOO) than predominant selfers (generally (1 70).

He proposed that this relation resulted

from less eficient pollen delivery for o w l e fertilization in cross-pollinating species associated with the uncertainties that accompany increased reliance on pollen vectors. Cruden suggested that P:O ratios evolved in the context o f a trade-off between minimizins pollen production to consewe resources and providing sufficient pollen to assure maximum seed set.

Cruden's explanation for the variation in P:O ratios has been criticized as over emphasizing pollen receipt by flowers (female function). and ignoring a hermaphroditic plant-s titness contribution as a male. through pollen donation (Charlesworth and Charleswonh 198 I; C harnov 1983). Instead. Chamov ( 1981) proposed that the optimal resolution of the conflict between investment in male and female hnctions occurs when the allocation pattern equalizes marginal fitness returns through both sex roles. From this perspective, an obligate selfing species should have relatively low male investment because an individual's pollen has access to the o\ules of only one female (itself), and fertilization does not involve competition with other males (Charlesworth and

9

Charlesworth 198 1 ). On the other hand. for outcrossing species the greater number of mating opportunities and increased competition for those opportunities favours production of many pollen grains to enhance an individual's competitive success as a male. C harnov's ( 1989) concept o f optimal resource allocation relies on the relations

between resource investment and titness gains through each ses hnction- The shapes of sex-speci tic sain relations depend on many aspects of reproduction. including resource availability. pollinator attraction. pattern of pollen export. post-pollination processes. inbreeding depression. seed dispersal and environmental conditions (Lloyd 1984: Lloyd 1985: Preston 1956: Ernms 1993). When female and male gain curves differ i n shape. selection favours Iess in\-estmentin the sexual hnction for which incremental fitness returns on investment diminish most rapidly with increasing investment. For animalpollinated. co-sexual plants siring ability may commonly eshibit strongly-diminishing returns (Charnot. 1 952; Lloyd 1 984: Harder and Thomson 1989: Harder and Wilson 1997). so that male hnction should receive fewer resources than female hnction (see

Figure 4.1). This differential resource allocation should be reflected in the P: 0 ratio. Widespread acceptance of C harnov's criticism of Cruden's interpretat ion of P:O ratios lead to a general neglect of the role of pollination in sex alIocation (although see

Lloyd 1 984); however. various esarnples imply that pollination characteristics influence

P:O. For instance. consider the unusual poll inat ion mechanism of Apprrticllrr he-wdro (Hydrocharitaceae) as described by Appert ( 1996). Male flo~versof this submerged dioecious plant. endemic to Madagascar. detach and float intact on the water surface. If a

male tlower drops into the meniscus created by an anchored female flower. pollen transfers directly. without the intervention of a po!len vector. This direct pollen transfer promotes intense competition among sibling pollen tbr access to the same ovules in a pistil (local mate competition). In association with these two factors. .-l. hr.-cnr~drcz apparently produces the fewest pollen grains per ovule of any outcrossing ansjosperm, 8. This example suggests that augmented et'fectiveness of pollen transfer. which causes

increased local mate competition, favours decreased relative investment in male hnction. Unusually low P:Oratios have also been recorded for outcrossing species that produce their pollen in aggregations. rather than as independent grains (Cruden and

Jensen 1979; Mehrhoff 1983: Broyles and Wyatt 1990; Morawetz and Waha 199 1; Neiland and IVilcock 1995: Nazarov and Gerlach 1997). Pollen-connecting threads are sugsested to resuIt in economic transport of aggregations of pollen in the Onagraceae (Cruden and Jensen 1979) and in some species of Annonaceae (klarawetz and Ciraha 199 I ). w-hich are also associated with unusually low P:O ratios Poltinia, characteristic of

the .Asclepidaceae and Orchidaceae. also facilitate movement of larse quantities of pollen between tlom-ers (Mehrhoff 1983: Broyles and Wyatt 1990; Nazarov and Gerlach 1997). Specifically. the coherence of the polliniurn limits the susceptibility of transported pollen to pollinator grooming. thereby reducing pollen loss durins removal. transport and deposition (Harder in press) (see sections 2.1.1. 2.1.2). -4s a consequence of efficient pollen removal and transfer, larse quantities of pollen from the same individual are deposited on a stigma. increasing the probability that all ovules in an ovary will be fertilized by a common maIe parent (Kress 198 1). This pollination pattern likely results

in severe local mate competition. because sibling pollen tubes compete among

themselves for access to owles. rather than with pollen tubes from other plants. 14

Objectives In this thesis, I present the results of studies on resource allocation in 1 7

temperate orchid species. The Orchidaceae is a highly distinct group. characterized by several reproductive attributes not present in other ansiosperms. This combination of features within one group invites investigation into the unique opponunities for resource allocation available to orchids. My primary objectives are ( 1) to evaluate the opponunities for adjustment of resource expenditure on gamete production and seed deveIopment during an individual reproductive season and ( 2 ) to investisate the relation betn-een pollination efficiency and the allocation of resources to the competing sex functions as indicated by the P:Oratio. To hcilirate interpretation of resource allocation in light of the impacts ofthe unusual traits that characterize the Orchidaceae, 1 summarize several reproductive structures unique to orchids which are crucial to my investigation in Chapter 2. In addition. Chapter 3 describes each species esamined with respect to general morphology. matins type. polliniurn type. reward production and pollinator behaviour.

In Chapter 3. I consider the opportunities for adjustment of maternal resource investment in the Orchidaceae. 1 measured fruit, owle and seed production, and seed

size in response to varying pollination treatments. The primary objective ofthis study is to determine whether orchids, unlike other angiosperms, adjust ovule production rather than seed production in response to pollen quality.

12 Chapter 4 considers resource allocation between the male and female sex functions. In particular, [ assess the relation between pollination efficiency and P:Oratio in orchids. M y interpretation of the results emphasizes the differences between autoyamous and xenogamous species and the effects of pollinium type o n the relation between P:O ratio and pollination eficiency. Finally in Chapter 5. I discuss my results from Chapters 3 and 4 by examining two patterns of innovation in the evolution of features characteristic to orchids and

consider the implications of these patterns to our understanding of orchid biology.

2

R E P R O D U C T I E FEATURES OF ORCHIDS A\D A DESCWTION OF

STLDY SPECIES 3.1

Unique Reproductive Traits of Orchids The Orchidaceae is one of the largest and most diverse families of plants.

including between 7- 10

?'ti

of all flowerins plant species (Dressier 1990). Several

reprociuctive features present in most orchid species distinguish this actively evolving

-group from other angiosperms. Some of these features, individuaIly or in combination, invite investigation into allocation of limited resources between and within competing ses fbnctions.

2.1.1

Column Animal-pollinated an~iosperrnsinfluence dispersaI of their pollen through the

duration and site of polIinator contact between the stigma and anthers by controlling pollinator position. stigma size and position, and anther size and position (Johnston 1991;

Harder and Barrett 1996). This mechanism has been refined in the Orchidaceae during the e\-olutionof the column, which maintains highly effective control over pollen removal and deposition (Burns-Balogh and Bernhardt 19%). The column forms by fusion of the stigmas, styles. staminode(s) and remaining

f.'unctional stamen (two stamens in the Cypripediodeae and Apostasioideae). In most cases these structures unite so completely that it is difficult to distinguish between them. Oniy tlvo stismatic lobes are fLnctional as the third forms the rostellurn, which separates the anther from the fertile stigmas on the column. In many orchids the rostellum breaks

14 down to f o m a glue-like substance. the viscidiurn. which helps attach the pollinia to the

pollinator. thereby reducin~grooming pollen loss during transpon (Dressler 1986). The anther is seated in a clearly defined area at or near the end o f t h e column, protected by the hood-like clinandrium (Proctor rr'a 1996).

1.I .2

Pollinia L'nfike most angiosperms. most orchids do not produce pollen as independent

erains. but package pollen in coherent structures called pollinia (van der Pijl and Dodson

C

1 966: Dressler 1 990). The development of pollinia. each of which contains enough

pollen to fertilize all the ovules in an ovary (Proctor and Harder 1994). represents a major innovation of orchid evolution (Dressler 1990: Arditti 1992; Wilson er cd. 1994). Pollinia ranse in coherence and complexity (Table 1. I); however; the evolutionary trend seems to lead towards development of harder. more compact poilinia. The package removed by the pollinator typically includes the pollinia, the viscidium and the stipe or caudicle. all of which comprise the pollinariurn. Some

pollinaria involve two units (hemipollinaria) derived fiom the anther, each with a separate L-iscidiumand poilinium. which may be removed either together or separately (Dressler 1990). In species with relativeIy compact pollinia, the caudicle (a softer

extension or tail) attaches the pollinium to the viscidium. Once pollinia attach to the pollinator. these caudicles function as both a 'stalk' and a weak point, permitting the pollinia to break away from the pollinator and remain on the stigma upon pollination (Dressler 1986). The caudicles produced by the anther shauld not be conksed with the st ipe. composed of non-sticky cellular tissue derived from the rostellum (Rasmussen

15

Table I . I

The relative abundance of pollinia types in the Orchidaeae (Dressier 1993).

Larger percentages are rounded to whole numbers.

Pollen / Poilinium Type

Percent of orchid species

Free monads

0.08

Sticky monads/tetrads

2.0

Mealy pollinia

6.6

Brittle pollinia

0.53

Sectile pollinia

11.0

Hard pollinia

80.0

I6 1956). which attaches the pollinia to the viscidium in most Vandoid orchids (Dressler

To promote cross-pollination, pollinaria of many orchids show characteristic movemenrs after being removed &om the flower (Danvin 1 562; Northern 1970). When pollinaria are removed from a tlower, they are positioned on the insect so as to strike the anther and not the stigma of another flower. Several minutes after pollinaria removal. differential d ~ i n and g bending or twisting of either stipe or caudicle changes the orientation of the pollen mass so that it will strike the stigma when the pollinator visits another flower. promoting cross-pollination

The Cypripediodeae and the Apostasioideae are the only orchid subfamilies that produce pollen grains as monads. Hotvever. Cypripediod pollen is sticky and paste-like due to the abundance of polIenkitt (Figure I . la; Dressler 1986). so that even though pollen is not packaged into pollinia. little pollen is lost through pollinator _nrooming. the typical fate of sranular pollen. AII remaining orchid species produce discrete pollinia that differ in the degree of

cohesion between the pollen _erains. Sofi powdery/mealy pollinia. uniform in structure and comprised of pollen tetrads. characterize most o f t he Spiranthoideae, and ancestral Epidendroideae (Figure I . 1b). Cohesion strands. extremely durable strap-like bands of sporopollenin. connect the four pollen grains that comprise a tetrad (Balogh 1982; Azkerman and Williams 19s 1;Fitzgerald rr

rl/.

1991). Sectile pollinia comprised of

aggregates of wedge-shaped massulae attached at their smaller ends to a central core of elastoviscin. are characteristic of the rest of the Spiranthoideae and Orchidoideae ( F i g r e

Figure 1 . 1 Examples of orchid pollinaria (pollinium plus viscidium). (a) Sticky, paste-

like pollen. (b) mealy pollinia (comprised of tetrads). (c) sectile pollinia and (d) hard waxy pollinia. .A. anther: F. filament: bl. massulae; P. pollen; Pn, polliniurn; V, viscidium

(%om Proctor and Harder 1994).

19

I . I c ) (Hesse and Burns-Baioy h 1984; Hesss, ef crL 1 989; Dressier I 993; Freudenstein and

Rasmussen 1997). The massulae vary in size and shape, depending on location within the pollinium and each contains many pollen grains (pers. obs.). Neither mealy nor sectiIe pollinia are deposited on the stigma of the visited flower in their entirety. Instead, pollinarion largely invoIves polliniurn fragmentation, which leaves several tetrads or massulae on each of several ditTerent stigmas, thereby pollinating a number of flowers (Danvin 1862: Dressler 1986: Xilsson 1992; Proctor and Harder 1994; Proctor c ? CII. ~ 1996: Nazarov and Gerlach 1997). The sectile condition is not intermediate between

mealy and hard pollinia. but rather represents a separate pattern of adaptation (Dressler 1990).

The most prevalent pollen packaging mechanism is the hard wasy polliniurn of t h e derived Epidendroideae (Figure I . Id; Knox and kIcConchie 1986). A coating of

smooth esine (sporopollenin) encapsulates only the outer tetrads of the pollinium. n.tlereas the inner pollinium tetrads completely tack exine and are surrounded by thin layer of intine (Yrung 1987b; Fitzgerald er a/. 1994). Pollinium integrity is maintained by an outer coating that covers the esine of the outer tetrads, forming solid connections

between tetrads and constructing a cuticle-like covering that physically links the outer tetrads into a structured pollinium (Fitzgerald rr a/.1994). Unlike other orchid pollinia, a pollinator removes a hard pollinium and deposits it on a stigma in its entirety. thereby

-rrrsatly eliminatins pollen loss during transfer. Only the outermost pollen grains of any type of pollinium have sporoderm. whereas the centrally located grains show progressive reduction of exine complexity (see

20 Pandolti and Pacini 1995; f-fesse

crl. 1989). This exine reduction, likely increases

-zerminability of the inner %rainsand avoids wasting material and energy for the production of an otherwise obtrusive and massive exine (Hesse rr a/- 1989). 3.1.3

Ovules Orchids are quite unusual in that the ovules are either weakly or not developed at

anthesis (Yeuns and Law 1997;.Arditti 1992: Zhang and O'Neill 1993). Pollination triggers ovule differentiation and pollen tubes must wait tbr ovule maturation untiI fertilization can occur (Law and Yeunp 1993; O'NeiIl rr CIZ. 1993). Ovule development apparently proceeds continuously once trigzered (Yeung pers. comm.; Nazarov and Gerlach 1997). This delay in ovule differenriation introduces an opportunity for maternal resource adjustment during h i t development which is not available in other angiosperms. because the number of ovules matured could be adjusted to the number of pollen grains received on the stigma. 3.1 .-I Seeds

Orchids produce multitudes of tiny dust-like seeds (0.15 - 6 mm) characterized by a loose, papery seed coat (with some exceptions: Dressler 1990). The absence of vascular connections between the ovary and owIes limits the potential for maternal resource investment into individual seeds by constraining nutrient supply (Rasmussen 1995; Yeung and Law 1997) and is associated with a lack of substantial nutritive tissue in

each seed. such as endosperm, for supporting seedling development (Benzig 1 95 1 ; Law and h u n g 1989; .Arditti 1992). Instead. orchid seeds must establish an obligate mycorhizal relationship for germination and early seedling growh (Dressler 1990; Arditti

21 1992; Rasrnussen 1995). Hence. the unique reproductive biology of orchids restricts

opponuniries for seed-size adjustment based on differential resource allocation between developing embryos which is available in most angiosperms. 2.2

Investigated AIberta Orchid Species I examined 17 species of orchids native to Alberta. Canada. with representatives

ofrhe four rypes of pollinia. includinz both self-pollinating (autogarnous) and crosspollinating (senogamous) species. 1 conducted all esperiments and collected all corresponding data during the summer (May to September) of 1996. Appendices 1 and 3 present the study locations and dates of esperimental manipulation. C)pripecliuni ctslceolus L. var. puhescens (W.) CorreIl (Yellow Lady's Slipper) and

C_r.pripediuttlptrsserirrrttfl Richardson (Sparrow's Egg Lady Slipper)

-

These sinnIe-flowered orchids produce sticky paste-like pollen composed of monads contained within two anthers (diandrous). one o n either side of the stigma. Uniike orchids with coherent pollinia. only a ponion of Opr-iperdirmt pollen is transferred

during a pollinator visit (Light and MacConaill 1998). The deceithl flowers of C~pr.iyrtli~lm cc~/ceo//ts offer no reward, but the bright yellow lip and somewhat fruity scent attract a variety of short-tongued bees (Halicitidae and Andrenidae), and possibly flies (Nilsson 1979; Catling 1985;Proctor el a/. 1996). A pollinator enters the lip through the obvious large opening. and becomes trapped inside. Escape from the slipper-shaped lip is possibIe oniy by one of two small exits at the base of the flower. As the pollinator forces its way out, the stisma provides support for the thoras (as it pushes down on the lip to make the tunnel wider) and any pollen on the

93

a-

thorax is deposited on the stigma. Further alonz the exit path, the pollinator is forced to push its thorax against the anther. where it picks up a new pollen load (Catling 1985). It is rare tbr a C'. ccrlcfiollw flower to receive its own pollen during a single visit, which in a

plant with few. large flowers should favour cross-pollination (Nilsson 1979). However, polIination of tlowers on the same genetic individual may occur. as all C~vpr~prtlirrms are rhizomatous and often grow in clonal patches (Kull 1 987; Proctor efnl. 1996; Currah rt

a/. 1986). Unlike In ('. cr[lceo/rfs.sel f-pollen transfer in CT. yc13~-e/-itrl1m is faci I itated bv a structural modification during development of the column. .As the flower matures, the anthers and stigma grow in direct contact due to shortening of the stigmatic branch (Catling 1985: Catling 1990). resulting in auto-pollination. This species Iikely relies entirely on self-pollination as a green leaf-like bract covers the opening to the lip of most

tlo~vers.effectively hindering animal poilinators from entering the flower. L.istertr horerrli-s3 ioronc - (Sonhern Ttv-ayb1ade) and I- cortkrtrr (L. ) R. B r. var. cortkrtct ( Heart-leaved

Twayblade)

The multiple-tlowered inflorescences of these small inconspicuous orchids bear from 5 to 15 pale white or g e e n flowers. Each flower contains a shallow groove down the center of the labellurn. which secretes nectar onto the surface and thereby attracts fungus gnats (klycetophylidae and Sciaridae). flies and small wasps (Nilsson 1981; Currah c.1 cil. 1986: Proctor el d 1996). Both Livlel-cr species produce soft mealy pollinia, composed of pollen grain tetrads (Dressier 1990). The Twayblade orchids have an intriguing mechanism to assist cross-pollination and prevent pollination between flowers

23

of the same plant (seitonosarny ). Before the pollinia are removed. access to the stigma is

blocked by the rostellurn. which lies directly below the stigma. Upon the gentlest touch o f a visiting insect. the rostellum explosively exudes a drop of viscid liquid. which upon

contact with the tips of the pollinia and the insect cements the pollinia firmIy to the insect's head. Immediately following pollinia removal, the rostellum bends sharply do~vn~vardsagain covering the stigma, however; rvithin 2-3 hours it straishtsns from its arched position. and exposes the stigma for the nelct visitor (Ackerman and klesler 1979: Proctor tit a/. 1996). Explosive ejection of the viscid matter From the rostelium may

startle t h e pollinating insect sufticiently to make it tly to another plant (Ackerman and YIesler 1979).

ti,.rrlw ha-errlis relies entirely on insect visitors to achieve pollination. whereas I_.

cof-~lLntr exhibits facultative autogamy by inactivating the trigger mechanism if polIinia

are not removed within several days of antheris (Ackerman and klesler 1979). Eventually the rostellum lifts. exposins the stigma to failins friable pollen from the incoherent pollinia above (Catling 1983; Catling 1985; Catling 1990). In north and west Europe L. cor~k~ra is Largely autogarnous (Proctor rr a/. 1996). .Tpircrnf/tes rorncmzofficmrr Cham. (Hooded Ladiess-tresses)

.As the name implies, Spircr)rrhrs produces an inflorescence composed of three

spiral ranks of creamy Lvhite. fragrant flowers. The pair of sot? mealy pollinia produced by .S~Y~I'CZIII~~..S flowers become attached to the tongue of visiting long-tongued bees (Hon~htt.~. ILdc)Iya&ifi~fc~e) and short-tonsued bees (Halictidae) (Catiin~1983; Proctor tct a/. 1996). In newly opened tlowers. the column lies close to the lip Ieaving onIy a narrow

24

passage for the tonsue of a visiting bee to reach the abundant nectar. In this position. the stigma is covered by the rostellum. which contacts the dorsal surface of the tongue of a probing insect (Carling 1983b). The viscidium with its attached pollinia becomes cemented to the upper side ofthe proboscis and is chereby removed from the flower by the poliinator. Two to four days follotving pollen removal. the remains of the roste1lum nither and the column and lip slowly move apart. esposins the stigma to pollen brought by a visiting bee from another tlower (Catling 1983b; Proctor rr a/. 1996). At peak

flowering. the lower. older flowers are hnctionally female whereas the upper younger tlowers are firnctionally male. thus bees arrive as pollen donors and leave as pollen receivers. increasing the likelihood of cross-pollination (Catling 1985).

-4~~ferorttlris rotunrlifo/i(r(Banks) Hulten (Small Round-Leaved Orchid) This small. dainty orchid produces a rather showy inflorescence of whitish, mau\-e-spotted tlowers. -4 pair of sectile pollinia attached to viscidia by slender caudicles originates immediately over the entrance to the nectar spur. A pouch-like rostelium encloses each viscidium.

visiting insect inserts its proboscis into the spur and touches

the rostellum. which bursts and exposes the sticky viscidia one or both of which touch the insect and stick firmly to it. The pollinia of Orchi,. species show characteristic

movements after removal from the flower (Danvin 1562). Initially the pollinia lie in the

same position as they occupied in the flower. In this position, if the insect were to visit another f-lower, the pollinia would be pushed against the poliinia there. However, about a minute afier their removal. the membrane forming the top of the viscidium dries, and each pollinium swings f o n ~ ~ a rthrough ds an angle of about 90". This movement brings

25 the pollinia into position to strike the sticky stigma ofthe next visited tlowers (Proctor er

(;Oodj?eruoh/ong~yi~/iff Raf var. obhng~Yo/irrCklenzies' Rattlesnake Plantain) and

ti- repens ( L. ) R. Br. var. repens (Dtvart- Rattlesnake Plantain) Evergreen basal rosenes giving rise to a large showy intlorescence composed of \c.hite t'lo~vers,characterize this senus of'lar~eIyclonal species (Kallunki 1981). Both I;oo~!t.rr-croh/ut~gIfcolicland (j.rtcyrtfi produce a pair of sectile pollinia attached to one

c*iscidium. Like Spirc~t~lhrs and Lisret-CIspecies. (joo~[t.rmexhibit protandry. Xnthesis begins at the bottom of the inflorescence so that during peak tlotvering, the Iower. older flowers are hnctionalIy female. whereas the upper younser flowers are hnctionally male. Protandry aiso occurs through movements of the column or both the column and lip (Ackerman 1975: Catling 19S3a). In young, male-phase flowers the position of the column (parallel to the lip) allows removal ofthe pollinium. but inhibits insertion of another pollinium and deposition ofpollen on the stigma. In older flowers, a slight raising of the column enlarges the space between the column and the lip so that a polliniurn can be inserted (Kipping 197 1; .Ackerman 1975; Catling 198Sa).

The abundant nectar and noticeable tloral odour of G. ohlcl~~gfidirr flowers attract

-

- bees (Bnmhr~sspp.), which visit the inflorescence from bottom pollinating long-tongued to top (Danvin 1562; KaIlunki 198 1 ). This pollinator behaviour combined with the slight protandry restrict, but do not prevent, geitonogarny (Kallunki 198 I ) . This species does

not self-pollinate auronomously (Kipping 1971 ).

26 Unlike G nh/o~tgIfc)/rn.G. reperts lacks a definite spur and floral odour. although it is also pollinated by Bomhtrs spp. Kallunki ( 198 1 ) sugsested that these two species

hybridize to form a distinctive intermediate tbrm named ti. rt?ssc?lntcr. Coe/ugfos.wntviride ( L . )Hanmann var. b r c t e u t t ~(Long-Bracted Orchid)

This common orchid forms a dense inflorescence of small yreenish flowers. The t ~ v osectile pollinia produced by CT(~e/og/o.vsr/fn attach to separate viscidia on either side

of a small stigma (Proctor et a/. 1996). Individual flowers secrete abundant nectar along a central ridge on the lip leading to a nectar-filled. sac-like spur. In addition. two small nectaries on either side of the lip beneath the viscidia attract srnaii bees. soldier beetles, and \\-asps (Catling 1 985: Willems and Mielser 1998). Like clrnt!~'otrl~is rorrit~difolin. I * ~ o / ~ ~ ~ y / o . v .pollinia s r / r n reorient following removal from the flower. The initial position of

the polliniurn on the insect prevents contact of the pollinium with the stigma of the nex? visited flower. However. 20-30 min afier removal the pollinium bends forward.

positioniny

it

such that in a subsequent flower it will contact t h e stigma rather than the

anther (Proctor rr a/. 1996). This long delay allows sufficient time for slow-moving pollinators to visit another spike.

P/(ttcrntlr~.rcr ~Iiintrrtr(Purs h ) Lindley var. ~Iiifrrcrta (Tall White Bog Orchid).

P. Itjperbore([ LL.) Lindley vat. Ii~pcrborc~r (Tall Leafy Green Orchid) and P. obtusnta (Banks ex Pursh) Lindley (Blunt-leaved orchid) These bog orchids produce multi-flowered inflorescences composed of small.

-greenish or white flowers characterized by a conspicuous nectar spur. Pi~rrclrtthercr

27

tlo\vers produce a pair of sectile pollinia, each with separate viscidia attached to the column just inside the nectary entrance (Kippins 197 1). The iarge. white-flowered intlorescence of P. di/ofarn produces a strong muskysweet odour at dusk and through the evening, which attracts noctuid moths (CatIing 1985). As the moths attempt to drain the nectar)., the viscidium attaches to the estended

proboscis. Unlike in other orchids. a recently opened P. Jilc7/c1/nflower positions its lip

such that it forces the insect to remove only one pollinium at a time (Catling and Caiiing 199 1 ). This unique mechanism may be advantageous to the plant in that it encourages

rhe next visitor to a flower to also act as a pollen disperser. The highly polymorphic P & L I / C Z/gperhort!~z ~ I ~ / ~ ~ ~ can L ~ self-pollinate autonornousIy by several mechanisms, which bring the pollinia onto the stigmatic surface (Currah e f nl. 19S6: Catling 1990). Self-pollination resuIts by falling of Friable pollen. falling or sliding

of entire pollinia. or bending of rhe caudicle eventualIy bringing the pollinia in contact u.ith the scipma. In addition. mosquitoes faciiitate cross-poIlination in this species

(Catling and Catling 199 1 ).

Pkrm/rl~tn.cr ohrrrscrta, the smallest of the three species. is polIinated extensively by mosquitoes (females ofrlrc/e.s spp.) and geometrid and pyralid moths. In all cases. the

pollinia attach to the insects' eyes (Stoutamire 1968; Thien 1969; Thien and Utech 1970; Gorham 1976; Voss and Riefner 1953; Catling 1983a). Cu!,.pso hu/bostr (L. ) Oakes var. rrrrtericcrncr (Fairy-Slipper Orchid)

This showy orchid produces a single fragrant flower, which attracts bees, but lacks a nectar reward. The two pairs of large hard. waxy pollinia are attached to one

28 viscidium. Ho~nhtf.sspp. queens pollinate this species by removing the entire pollinarium

(both pair of pollinia and viscidium), which becomes attached to the scutelIum on the upper posterior surhce of the thorax (Mosquin 1970; Stoutamire 1971; Boyden 1982). As a bee lands on the tlower, it pushes its head and thorax in under the column- As the thorax is brought back out, it touches first the stigma. depositing individual entire poliinia (when availabie), then the anther. picking up a new pollinarium (Catling 1985: Proctor and Harder 1991: Alexandersson and Agren 1996). .Although it has been suggested that

the circular patches of C'c~[vy.sorepresent vegetative growth (Mosquin 1970; Boyden 1593). Currah and colleagues ( 1986) could not substantiate this claim and suggested that

individual plants in patches develop from seeds. Ci,mllorlti.:rr rrrtrcukrtcr Raf. var. nracuI41ti (Spotted Corai-root ). C. strirrtn Lindley (Striped Coral-root) and C tn'fr(i(1ChateIain (Early Coral-root) The saprophytic coral-roots produce c o l o h l inflorescences of pink-striped. pink-

spotted or greenish-yellow flowers. A tlowx- produces two pairs of hard. round pollinia, each attached to the column by a stipe.

Predaceous dance tl ies (Enrpiclicr'nr) poi l inate Cbr.n//or-irizo n~acrrlorn. When a fly probes the nectar): entrance. it brings its notum directly in contact with the column. As the fly withdraws its proboscis, its thora.. pushes against the end of the column and contacts the viscidiurn withdrawing one or both pairs of poI1inia (Kipping 1971). If

insect pollination is not efi'ected, C: N~CICIIICIILT exhibits delayed self-poIlination (Catling 19S3a: Catling 1990). The anther cap in an unvisited flower dries and falls off. exposing

29

the poliinia, which rotate around the edge of the rosteIlum and onto the stigma even though they remain attached to the column (Kipping 1971). The lip of C'orrrIlorhkastric~(a is not lobed and does not contain a spur. However it is likely that flies also pollinate this species.

In contrast to the other coral-roots, the greenish stems and tlowers of CWornI/orhiza r/-!f7dn may photosynthesize to a limited extent (Catling and Catling 1990). Like C: tnclmlcrrcr, this species exhibits autonomous self-pollination by rotation of pollinia onto

t h e stigma (Catling 1983a: Catling 1990);however; this mechanism occurs prior to complete anthesis (pers. obsewation), so that this species probably cannot cross-pollinate.

;t/crk(~vis riinnoplt_).llos(L.) S wartz

This tiny orchid produces a multi-flowered inflorescence composed of minute green-yellow flowers. Each flower contains nvo pairs of hard pollinia attached to two

C

ciscidia. Almost nothing is known about the pollination of this species; however, .L/cr/c~vismay self-pollinate autonomously by rotation of polIinia onto the stigma (CatIins 1990).

30 3

OPPORTLXITIES FOR .ADJL~STbLENTOF bIATEkVrU, RESOLrRCE

LXVXSTMENT IN THE 0RCHIDACE.A.E

3 1. I

Adjustment of Maternal Resource Investment Maternal plants have t h e opportunity to adjust resource expenditure on gamete

production and seed development at several stages during a reproductive season (see Lloyd 1980: Lloyd er al. 1980; El Kablawy and Lovett-Doust 1996). including ( 1 ) floral initiation (number and location of flowers produced). (2) abonion o r development of each fruit. and ( 3 ) alteration of both final seed number and size through variable allocation of resources among retained ovaries within a Fruit (Harper 1977). Except possibly during ultimate allocation to seed size. resource regulation afier flower initiation involves a decrease in resource allotment (downward adjustment). Unlike most angiosperms. orchids possess several traits that allow unique opportunities to adjust maternal investment. .At anthesis the immature orchid ovary lacks mature owles and ovule development is suspended until stimulated by pollination (b'ithner er tri. 1974; .Arditti 1993; Zhang and O'Neill 1993; Rasmussen 1995). This delay in ovule development could enable increases or decreases in the number ofot-ules matured

to match the number of pollen grains received by the stigma, creating an opportunity for resource consewation or redistribution (Yeung and Law 1 989; Ardi tti 1 992; Rasmussen 1995; Nazarov and Gerlach 1997). Consequently. maternal investment could be adjusted

31

upwardly as t h e resources put into o d e production could be increased to match a high pollen load. Other features of orchid poIlination may also faciIitate characteristic patterns of maternal resource adjustment. Because orchid pollen is transported in poilinia which conrain sufficient pollen to fenilize most ovules in an ovary (Arditti 1992), one pollinator \visit can induce h i t maturation. resuIting in production of many seeds (Proctor and

Harder 1994: Nazarov and Gerlach 1997). ConsequentIp, even if adjustment of ovule number does not occur. such high pollen receipt should result in high proportional seed set in pollinated tlorvers. Pollination of orchids with pollinia results in high genetic similarity of offsprin,o, so the maternal parent may benefit more from controlling resource investment in seeds by whole-hit abortion than by terminating individual e m b ~ o (Kress s 198 1 ). A s multiple h i ~ provide s alternate opportunities tbr reproductive success. this mechanism of resource consemation should occur more commonly in plants with multiple-flowered- rather than in single-flo~vered.inflorescences. In addition. orchids may also adjust resource investment during seed production. Orchid capsules with mature seeds often also contain otules with

failed embryos (Catiing 1982; Schmidt and Antlfinger 1992). which could result from

inadequate pollination, unsuccessfU1 fertilization or embryo abortion. Unlike most angiosperms. orchids may exhibit limited opportunities for seed sizenumber adjustment. The small size of orchid seeds, their apparent lack of nutritive

-

provisionins (Law and Yeung 1989; Arditti 1992), and obligate reliance on mycorhizae

for gemination (.kditti 1 992; Rasmussen 1 995) may preclude increased or decreased

maternal investment in each seed. Consequently. pollen quality should not influence seed size within orchid species. 3 I 2

Objectives This study assesses the regulation of maternal investment by orchids in response to

hand self-, hand cross- and natural-pollination for 17 temperate orchid species. The

primav objective considered whether orchids. unlike other angiosperms. adjust ovule

production rather than seed production in response to pollen quality. I addressed this objective by examinins the effects of self-, cross- and natural pollination on all stages of maternal post-pollination investment: fmit set. otvle deveIopment, seed set and seed

development. I considered response differences to pollination treatments beween species of opposing mating types (autogarnous vs. xenogarnous) within a genus and between

-zenera.

I did not esamine responses to differences in pollen quantity explicitly: however

comparisons of natural pollination with hand-pollination should reveal some quantity effects. as a natural pollen load may often deliver somewhat less than I applied during hand pollination. 3.2

,CIETHODS AND bI.ATERI.US

3.2.1 Study Speciesandsites

This study considers ovule and seed production by 17 species of temperate orchids in response to dicerent pollination treatments (two species were not subjected to

33

hand-pollination). Some of these species are circumpolar, whereas others occur only in North America. All study sites were in .iUberta. Canada. Based on previous studies, 1 chose the species to represent the four pollinia types (sticky, mealy. sectile and hard) and two mating types (autogamous and senogamous) found in orchid genera in Alberta. I conducted the experiments and collected all corresponding data during the summer (May to September) of 1996. Appendices 1 and 2 present the study locality and dates of pollination and seed capsule collection of each species. 3 -9.2 Pollination Treatments

The pollination treatments that a species experienced depended on whether it was reported to be autogamous or xenogamous (see section 2.2). Species purported to be autogarnous were not esposed to hand-pollination. To confirm that these species set seed autonomously. a third of the individuals examined were bagged when the flowers were in bud until capsule formation began (autonomous self-pollination). The remaining plants were not bagged and

so couId have received pollinator visits (natural pollination).

Most xenoyamous species were subjected to four pollination treatments. Three treatments involved basging to exclude pollinators. To confirm the mating capability of each species, one randomly selected third of the bagsed piants did not experience handpollination (autonomous self-pollination). To assess the self-compatibility of each species,

I hand-poIlinated flowers o f a second third of bagged plants with their own pollen and removed the remaining pollinia to prevent autonomous self-pollination (hand selfpollination). The hand cross-pollination treatment involved removal of a plant's own

pollinia and hand pollination with pollinia from an individual at least 1 m away. The founh. unbasged group of plants experienced natural pollination. These latter plants were not identified until fruit set so t did not assess the proportion of flowers setting fruit for naturally pollinated plants. Plcrmnlhtirn ~ii/oram and -L.(alcruisrnot?oph,t'~~os were not predetermined study species and were encountered by chance. I located these species during peak flowerins and collected only naturallj- pollinated fruit fiom them. I could not determine from the literature the mating system of either of these species; however, the relatively low seed set of P. clilnrora (see section 3.3.3). along with the strong fragrance of the large showy inflorescence suggest xenogarny. On the other hand, the high proportional seed set (see section 3 . 3 . 3 ) and diminutive flowers of iLI. mor~oph_t.llos.suggest autogamy. Each hand-pollination treatment involved from seven to 23 plants, depending on availability. For species that produce more than one flower per plant, three to five adjacent flowers on an inflorescence were chosen for manipulation. Each hand-pollinated flo~verreceived a pollen load equal to half the pollen produced by an individual flower. Ctp-lpedrm?flowers produce two sticky masses of pollen. rather than discrete pollinia

which precluded accurate determination of pollen load. Consequently, using a spatula I removed a smear of sticky pollen from one of the anthers and pollinated Cypt-ipedfiin~ flowers by spreading the removed pollen on the stigmatic surface. -41flowers were bagged following pollination to prevent insect visitation. Once seed capsules began to

35

develop. the plants were unbagged. Seed capsules from hand- and naturally pollinated plants were collected several weeks later when they were hily mature. These methods had to be modified slightly for Listera boredis and Spirm~rhes I ~ ~ ~ c ~ ~ I = o which ~ $ L ~ exhibit J I c I .protandry

so that the stigma does not become receptive until

a few hours to days after pollinia removal (Catling 1985: Proctor et a/- 1996). For these species, I removed and deposited pollinia on consecutive days. rather than simuItaneously. 32 . 3

Otule and Seed Counting 1 examined approximately 20 capsules per species to quantity ovule and seed

production per flower. rUthough in some cases several seed capsules were collected from the same plant, I used only one randomly chosen capsule per plant. The capsules were chosen to represent pollination treatments equally. The tiny, dry seeds of orchids are easily disturbed by air currents and static electricity, so I modified the method of Proctor and Harder (1994) to prepare ovules and seeds for counting. To suspend the ocvles and seeds in liquid, 1 dissected each capsule in a 15 mL jar by splitting it along the midvein of each carpel and scraping the carpel wall

with forceps to free all otuies and seeds. To remove any remaining seeds, I washed the empty capsule with water and then checked all pieces under a dissecting microscope. I then increased the volume of water to 15 mL, added a few drops of surfactant (1% Tween), and partially screwed on the jar lid, allowing air exchange but preventing over-

bubbling of the suspension. The jars were then subjected to a pressure of -90 kPa in a vacuum oven for 15 min to suspend the seeds in solution by forcing the air out of each

36

seed. To facilitate viewiny of the otherwise colorless ovules and seeds I added a few drops of acid hscin dye to each jar. Rather than count all the ovules and seeds produced by a capsule, I counted a 15%

sub-sample. 1 transferred the stained suspension into a larger container and increased the

volume to 100 mL by adding water. While a magnetic stirrer kept the seeds evenly suspended without vonexing. I pipetted three 5-rnL sub-samples of the solution into small petri dishes. I counted all ovules and seeds in each sub-sample at I2x under a dissecting microscope. 1 distinguished seeds from owIes by the presence of a stained, round. dark

embrq-o. which was clearly visible through the transparent seed coat. I quantified seed

proportion as the ratio of seeds to total ovules per capsule. For three seeds per subsample I measured the length (L) and width (W) of the seed coat under a dissecting microscope at 50s. I used the mean seed volume (RLW'/~)per flower in further analyses. Ocxles were not measured. 32 - 4

Owle Fate The failure of all ovules to develop into seeds could resuIt from 'overproduction'

ofokules and/or fertilization failure / embryo loss. Using the data on ovule fates (section

3 . 2 . 3 ) . I determined proportional ovule overproduction per fruit (POI.) as

where o is total ovule production and s is seed production.

37 distinguished between ovule

'overproduction' in naturally self- and hand self-pollinated flowers. and hand crosspollinated tlowers. To quantify the proponion of owies that failed to be fertilized and/or failed after fertiiization (Pffl. 1 used the ratio of the difference in mean seed set by hand cross-pollinated (s;) and hand self-pollinated (s,j tlowers to s:.

Because proportional fertilization failure and/or embryo loss is calculated from

species means, rather than per flower, I used the jacknife procedure outlined in Sokal and Rohlf ( 1905) to determine standard errors. This procedure involved randomly exc1uding one obsemation for the estimates of seed number from hand cross- and self-pollinated flowers and calculating Pffusing the remaining observations. I repeated this process untiI each obsen-ation had been excluded at least once. These estimates of Pffwere then used to calculate a mean and standard error

3 . 2 . 5 Statistical .Analyses

I used general linear models (Neter rr a/. 1985 : GLM procedure, SAS 6.12) to assess variation in the dependent variable. Analyses that involved continuous covariates in addition to categorical main effects initially considered all possible interactions involving covariates. Non-significant effects involving covariates were excluded by bach~vard A posteriori contrasts involved Tukey's studentized range for main elimination (u=0.05).

effects and ~ i d a k ' smultiplicative inequality for interactions to control the experimentwise

38

Type I error rate to u=0.05(Sokal and Rohlf 1995). The binomial nature of proportional fruit and seed set required losistic transformation to linearize the response fincrion and weighted least squares to account for the unequal error variances meter rr ni. 1985). Ovule production and seed volume were square-root transformed. To facilitate

presentation. I back-transformed descriptive statistics. resultins in asymmetric standard errors. which I report as lower (LSE) and upper standard errors (USE). I assessed variation between poHination treatments in ( I ) the proportion of flowers

that set fruit within each species and among genera and (2) the proponion of owles that matured into seeds to quantify seed production within species. The invariable I0094 h i t set of the three single-flowered species (C,c'yrrpedirirn pnsserrrri~m.C. cnlcro/iis and .sternsascertained prior to experimentation. I examined the effects of pollination treatment on ovule number and seed volume

within species. I also examined whether variation in polIination treatment. ovule production and seed proportion explained significant proportions of the variation in seed volume wirhin species. 3.3

RESULTS

3.3.1 Otule Production

Ovule number per flower varied over 60-fold between species, ranging from a few hundred (Li.sterrrcor~faro) to tens o f t housands (e-g..Cnopso bcilboscr: Table 3.1 ). Within all but one species. ovule production did not differ among pollination treatments (Table 3.2). In the case of L. h~rc't-rlis.self-polIinated flowers produced about half as many

o~,ules as outcrossed flowers. but hand outcrossed and naturally pollinated flowers

produced equivalent numbers of ovules (Table 3.1 ). Pfatarrrhercrrlikrtclta and !Lfcr/mis r ~ l o ~ ~ c ~ p / [were v / / u snot exposed to any pollination treatments because they were not

located until after peak flowering.

I examined the combined effects of mating type and pollination treatment on

ovule production by comparing congeneric pairs of species with contrasting mating systems (Table 3 . 3 ) . The results of this analysis depended on the inclusion of C)p;'ytcr/ilrm species. M'hen C>priyrclirtnt was included in the analysis, pollination treatment significantly affected

40 Table 3 . 1 Ovule production (x lo-') (mean. LSE-USE,11) in response to four pollination treatments. See Table 3 . 2 for results o f overall statistical analyses. Superscripts a and b denote differences among means within a species based on Tukey's studentized range, Pc0.05 (Sokal and Rohlf 1995).

Species

Autonomous self-pollination

Hand self-poltination

Hand crosspollination

hratural pollination

Sticky pollinia

Mealy pollinia

SectiIe pollinia

-

--

-

continued ...

41

Table 3.1 continued.

Species (;ooi&el-n

Hard pollinia

.Autonomous self-pollination 0

Hand self-pollination 2.2

and crosspollination 2.5

Natural pollination 1.9

42

Table 3 . 2 Analysis of the effects of pollination treatments (natural, hand cross-

pollinated and hand or autonomous self-pollinated) on ovule production. proportional seed set and seed volume. The analyses of ovule production and seed volume considered square-root transformed data, whereas those for seed proponion involved

logistic transformation.

Species

Pollination Treatment Otvfe Production Seed Proportion

Seed Volume

continued..

Table 3.2 continued. . -

Species

Pollination Treatment Ovule Production Seed Proportion

Seed Volume

p-~p

1

This autogarnous species was not hand-pollinated. hence it was subjected to only two

pollination treatments (natural and autonomous self-pollinated).

* 0.05>P>O.O I, ** 0.0l>P>O.OO 1. *** P0.25j. Ovule production also differed significantly among species within zenera (Table 3.3). Autogarnous species produced more owles in Piclictt~rhrm( t 1 2 5 = 2.86, Pc0.05).C)prcpedilrm ( I , : ~= 5.30, P= 4.13 *

F6.J60 =

1 . 12

F I . J=~4.8i *

Fl.160

Pollination Treatment

s Genus Species (genus)

= 0.07

Figure 3 . 3 The effects of experimental pollination treatments on mean ( 5 9504 confidence interval) proponional h i t set in (a) autogarnous and (b) xenogarnous orchids. The plants in panel (a) were bagged. but not hand pollinated. Those in panel (b) experienced either

hand cross-pollination (solid symbols) or hand self-pollination (open symbols). Only one

species was involved in each genus per panel. except for autogarnous Cora/lorhiza, and senogarnous Gour!sero. which involved two species each. See Table 3.2 for associated statistical analyses.

57

the multiple-flowered species overlap the 100% fruit set realized by the single-flowered species. Cn&p.so httlhosa and C:rpripeciirrm cnlcrolrts. Pollination treatment significantly affected fruit set by only I.. h o r d i s and C. l9iridr(Figure 3.3b). Hand self-pollinated flo~versof L. horeolis set significantly more f i i t than hand cross-pollinated flowers (/l60

=

2.8 1. PI.. L.Plitmann and C.C. Heyn. 1991. Reproductive effort in desert versus Mediterranean crucifers: the aIlogarnous Erztcnrin msrrcrm and E. hispmriccr and the autogarnous Erophilcr mirlirncr. Oecologia 100:256-292. Borba. E.L. and J. Semir. 1998. Wind-assisted tly pollination in three Bufboph~.lllrm (Orchidaceae) species occurring in the Brazillian Campos Rupestres. Lindleyana 13:303-2 IS.

Boyden. T. 1952. The pollination biologlr of CTcl[rpsobrtlhosa var atnericmxz (Orchidaceae): initial deception of bumblebee visitors. Oecologia 5 5 : 175-154. Broyles. S.B. and R. Wyatt. 1990. Paternity analysis in a natural population ofrlscfepiczs e-~-crlrcria:multiple paternity. hnctional gender. and the 'pollen-donation'

hypothesis." EvoIution 44: 1454- 1465. Broyles. S .B. and R. Wyatt. 1995. X reexamination of the poilen-donation hypothesis in an experimental population of Asclepim e-rolfc~fa. Evolution 49: 89-99.

Brunet. J. 1991. Sex allocation in hermaphroditic plants. Trends in Ecolo_syand Evolution 7:79-51. Brunet. 3. 1996. Male reproductive success and variation in fruit and seed set in

.-Iqrti/