, .

Journal ofVegetationScience13: 885-892,2002

A

@lA VS,'OpulusPressUppsala,

885

~

: ..

-r

.,¡ '-'

FORUM Reproductive successin fragmented habitats: do compatibility systemsand pollination specializationmatter? Aizen, Marcelo A.l; Ashworth, Lorena2 & Galetto, Leonardo2* I Laboratorio Ecotono, Universidad Nacional del Comahue,Centro Regional Bariloche, Quintral1250, 8400 San Carlos de Bariloche, R{o Negro, Argentina; 2Instituto Multidisciplinario de Biolog{a Vegetal (Universidad Nacional de Córdoba - CONICET), C,C, 495, 5000 Córdoba, Argentina; 'Corresponding author: Fax +543514332104; E-mailleo@imbiv,unc,edu,ar ":-

Abstract. This paperexploreswhetherplant breedingsystem and pollination specializationinfluence fue reproductiveresponseof plantsto habitatfragmentation,It is meaningfulfor conservationto predict a plant species'extinction risk, We found 25 studies in fue literature assessingfue effects of h 11 d ' fr habltat agmentatlon on elt er po matlon or repro uctlve successof 46 plant speciesto answerthe following questions: 1, Are pollinationandreproductivesuccess of self-incompatible speciesmorelikely to declinewith habitatfragmentatiopthan the pollination and reproductivesuccessof self-compatible species?Although most of the speciesshowedstatistically significant negativeeffects,fue pollination and reproduction of self-inc~mpatiblespecieswere as li~ely to d~cline with fragmentatlonas those of self-compatlblespecles,2, Are pollination andreproductivesuccessof specialistplantsmore '

o

,

,

0'

o

,

o

affectedthanthe pollInatlonandreproductlon of generallst

plants?Comparisonsof fragmentation-related changesin pollination and reproductive successbetween specialistsand generalistsdo not supportfuehypothesisthat specializationin pollination increasesfue risk of plant extinctiono3, Can selfincompatiblespeciesoffsettheir expectedhighervulnerability to fragmentation by being, on average, more pollination generalistthanself-compatiblespecies?In a largerdataseton 260 species,we did not find significant differencesin either , o , , o fue mean number or frequency distribution of numbersof flower-VlSltlngspeCles or ordersbetweenself-compatibleand self-incompatiblespecies,OUTreview suggeststhat no generalizations can be made on susceptibility to fragmentation basedon compatibility systemandpollination specialization, ,

o

o

Keywords: Breedmg system; Habltat fragmentatlon;speclallzatlon;Reproductlon;Self-compatible;Self-mcompatlble, o,

o

,

,'o

.

Introduction It is largely acceptedthat habitatfragmentationhas distinctiveeffectson plant andanimalpopulations.Particularly, a r~ductionin populationsizeandan increase in isolation related to fragmentation mar be linked to

;

-

:~ ;

increased inbreeding, decreased individual fitness, fue los s of genetic variation, and consequently to increased risk of population extinction (Murcia 1995; Jules & Rathcke 1999' Jacquemyn et al, 2002 and citations th ' ) I d'd ' t ' t ' t d' erem,

n a

I10n

,

o 1 s

'o

lTec

t

consequences,

f rag-

o

mentation m~y alter mteractio~s among speCles(Kattan & Alvarez-Lopez 1996; MurCIa 1996), More than 80 % of fue extant flowering plants depend, to different degrees, on ánimals for their pollination and sexual reproduction (Bawa 1990; Buchmann & Nabhan 1996), Although the evolutionary acquisition of animals as pollen vectors boosted pollen transfer efficiency, dependence on mutualists for reproduction could have increased 1 t ti' b' l ' t t f t t' p an suscep 1 1 Y O ragmen a Ion and other "lorms of " , habrt~t dlst~bance (e,g, B~nd 19?4; Spl~a200 1), Disruption of plant-poll1nator mteractions can occur becauseof fue sensitivity of many flower visitors to fue changes in habitat quantity and quality triggered by fragmentation (Kearns et al, 1998; Aizen & Feinsinger in press). As a result, fue degradation of this plantanimal mutualism mar provoke sizeable decreases in seed number and quality and even constitute the first step t owards th e demograp ' hlC ' co11apseof many p1ant , populations, Th,ere,area few examples of plant laxa on the verge of extinction due to a lack of sexual reproduction associatedwith fue los s of their original pollinators (e.g, Buchmann & Nabhan 1996; Renner 1998; Cox & Elmqvist 2000; Paton 2000). However, plant species ' l' t ' tat fra ' fti can d1 er m thelT vu 1fierab11 y to hab1 gmentati' on accordmg to th elT ' d ependence on po11ma ' ti, on mutual Ism, ' ' '

---:;

1

'

In particular,two reproductivetraits could be important in determiningfuedegreeof reproductiveresponsiveness of plantsto habitatfragmentation:breedingsystemand pollination specialization(Bond 1994; Murcia 1996; Renner1988;Aizen & Feinsingerin press), Plantbreedingsystemsrangefrom thosethatenforce

'.,

1

. outbreeding to those that ensure sexual reproduction

I ...

.

i !

\~i7Z *' r"::

~

886

Aizen, M.A. et al. ...

via autonomous,within-flower selfing andautogamous seed set (Lloyd 1992). Among the most common outbreedersareplant specieswith distinctive mateand female individuals (i.e. dioecy), and thosewhich producehermaphroditicflowers but possessa geneticallybasedself-incompatibility system.On fue other hand, fueTearemanyself-compatiblespeciesthat cansetseed via selfing. In animal-pollinatedspecies,this inbreeding-outbreedinggradientwill dictate,beyondits genetic consequences for plant populationsand individual fitness,the ü.veralldegreeof dependenceon the pollination mutualism for plant reproduction (Bond 1994). Whereasreproductionof obligate outbreederswill requiTefue presenceof othermatesanda palIen vectorto transferpalIenbetweenthem,reproductionof inbreeders will bemostly independentof fuepresenceof matesand animal mutualists. Plants can also vary in their degreeof pollination specialization, from extreme specialists to extreme generalists.Pollination specialistsaredefinedasplants pollinatedby one or a few ecologically similar animal species,whereasgeneralistsare plants pollinated by severalto many species,usually of diversetaxonomic origin (Renner 1998).The yucca/yucca~moth and fig/ fig-wasp mutualismsarewell-studiedcasesof extre,tne specialization.Rowever, flowers of most speciesare usually visited - and presumably pollinated

- by animal

visitorsthatcanvaryfrom a few, to morethan100species (Feinsinger1983;Rerrera1988;Waseret al. 1996).This gradientin pollination specializationmay be relatedto the likelihood of mutualismfailure. Pollinationspecialists areexpectedto be more vulnerablethan generalists becausefue loss of only one pollinator could lead to a completereproductivefailure (Bond 1994). Recentreviews (Bond 1994;Murcia 1996;Renner 1998)describein detail how dependenceon the presenceof other plant individuals andon particularpalIen vectorscan make someplant speciesmore susceptible than others. Rowever, mainly becauseof the lack of much published data, there has not been so far any formal testing of fue hypothesesthat breedingsystem andpollination specializationdeterminesa species'reproductiveresponseto habitatfragmentation.Someauthorshavesuggestedthat single-variablepatternsmight not be detectedbecauseplants could exhibit a suite of compensatoryreproductivetraits that make them, on average,equally resilient (or susceptible)to the effects of habitatfragmentation(Bond 1994;Jules& Rathcke 1999).Particularly,Bond (1994) arguedthat extinction may havealreadyremovedhigh-risk specialistswith no backupreproductivemechanismsandthat self-pollination may haveevolvedrapidly in responseto pollinator failure. Rere,we try to answerfue following questions: l. Are fue pollination andreproductivesuccessof self-

incompatiblespecies(i.e. obligate outbreeders)more likely to decline with habitat fragmentationthan the pollination andreproductivesuccessofself-compatible species(i.e. facultativeinbreeders)?2. Are fue pollination and reproductivesuccessof specialistplantsmore affectedby fragmentationthan fue pollination and reproductionof generalistplants?3. Can self-incompatible speciesoffset their expectedhighervulnerability to fragmentationby being, on average,more pollinationgeneralistthan self-compatiblespecies?

Data sets .

We havebuilt a databaseof> 10000 referenceson plant reproductive ecologyfrom fue 1991-2001Current Contentsdatabase.Accordingto the threeobjectivesof this study we selected406 papers,out of which 25 studiesassessed, either explicitly or implicitly, the effects of habitat fragmentationon either pollination or . reproductivesuccessof 46 species.Someof thesestudies ass~sspollination and reproductionof (1) plants in true habitat fragments(e.g. Aizen & Feinsinger1994; Cunningham2000); (2) isolated trees in pasturesvs. thosein forests(e.g.Aldrich & Ramrick 1998;Rocha& Aguilar 2001); (3) plant patchesof different sizes or degreeofisolation (e.g. Morgan 1999; Steffan-Dewenter

& Tscharntke 1999). We included the third type of studiesbecausemanyof fuemechanismsinvolvedin the so-called'fragmentationeffects' arepopulation-sizeor -isolationdependent. For eachspecies,we compiledinformationon family, location (latitudinal region and continent),habitat type, growth form, breedingsystem,andtype offlower visitors. All 46 specieshave hermaphroditicflowers. All studiesbut one (Cunningham2000) provide information on whetherthe focal specieswere self-compatible (SC) or self-incompatible(SI). For self-compatible speciesand if available,we also computedeither fue percentfmit or seedsetwhenpollinatorswereexcluded (i.e. capacityfor autonomousself-pollination).We classified each speciesas a pollination-specialist(S) or pollination-generalist(G) basedon the taxonomicarray of flewer visitorsmentionedin fuereference,pollination mechanism(e.g.buzzpollination), andflower morphology.Althoughtheremaybea subjectivejudgementin fue assignmentof someplant speciesto eithercategory,we believe that our dichotomousclassificationcapturesa large part of fue variation in fue specialization-generalizationgradientcharacterizingthis speciessample. We alsoincludethe effectof fragmentationon pollination and reproductive success.Although many of fuesestudiesprovidequantitativeestimatesoffragmentation effects, we consideronly fue qualitative effect

t

successin fragmented habitats -

- Reproductive

887

'$

"

(negative, neutral, positive) because fue range in fragment size and degree of isolation included in these studies were so dissimilar that estimations ofthe magnitude of fuese effects were meaningless for comparative purposes. For many species, pollination levels were estimated by counting pollen tubes (Aizen & Feinsinger 1994) or through pollen limitation assays(e.g. Jennersten 1988; Moody-Weis & Heywood 2001). These measures could reflect either quantitative (i.e. related to visit frequency, pollinator efficiency and capacity for withinflower self pollen deposition) or qualitative aspects of pollination (i.e. determined by breeding system and pollen transfer), However, in a few studies only quantitative estimates of pollination were provided through measuring either pollen deposition (e.g, Rocha & Aguilar 2001) or pollinator visitation frequency (Smith-Ramírez & Armesto in press). In most studies,reproductive success was estimated as fruit set, seed set or total seed output, In a few studies, however, other aspectsof reproductive success such as seed germination (e,g. Menges 1991; . Ouborg & van Treuren 1995) were considered. Information on fue species,habitats,and qualitative fragmentation effects is summarized in Table 1. To address our third question, we also gathered information, on breeding system and pollination specialization of 260 species, from 183 genera and 85 families, This data set (available from the authors on request) includes species that represent most existing growth forms and occur in a widerange of tropical and temperate habitats in all five continents. Based on the qualitative or quantitative information provided, we assigned each species to one of three breeding system categories (dioecious, self-compatible or self-incompatible). We considered fue number of either flowerd d +" h art 1 . .. , VlSltlng specles or or ers reporte 10r eac p lCUar plant species as a measure of pollination specialization.

lack of difference between SC and SI speciesis surprising. A potential source of bias that might influence the similar response of SC and SI species is that studies dealing with effects of fragmentation (or patch size) on plant reproduction often focuses on species with showy flowers that depend on pollinators for seed set, regardless of their breeding system. Thus, the pollination and reproduction of fue self-compatible species included in fuese studies could be as mutualist-dependent as those of most obligate outbreeders. This subset of self -compatible species, exhibiting a mixed mating system, also express inbreeding depression at early stages of reproduction (e.g. during seed set; Klekowski 1988),particularly in small populations trapped i~ habitat fragments, However, negative responsesof fragmentation at either pollination or seed set stages were also found in selfcompatible specieswith a large capacity to produce seed autonomously (e,g, Gentianella germanica, Nepeta cataría, Portulacá umbraticola; Table 1). Considering the available evidence, we can conclude that some selfing capacitydQesnotlessentheprobabilityofaplantspecies to respond negatively to fragmentation, Given the lack of information of the compatibility system for a few species (Table 1) and potential wrong assessmentsfor others, we also used variation in growth form as a surrogate ofthe outbreeding-inbreeding gradient with trees occupying the outbreeding extreme and herbaceous plants the inbreeding extreme (Klekowski 1988). According to Murcia (1996) trees would be the group most susceptible to forest fragmentation, not only due to a high incidence of self-incompatibility but also because of low density.

-

;"'.

o

¡::::]

100

Habitat fragmentationand breedingsystem Are the pollination and reproductive successof self -

-

Self-compatlble Self-incompatible

P =0.4314

80

~ ~

[)'

60

g

40

incompatiblespecies(i.e. obligateoutbreeders) more

ffi

compatible species(i.e. facultative inbreeders)?Ca, 70% and 60% of the species listed in Table 1 experienced statistically significant negative effects on pollination and reproductive success, respectively, However, fue pollination and reproduction of self-incompatible species l'kl d l. 'hf . h f were as 1 e,y to ec .me w,Jt ragmentatlon as t ose o

Il: 20

Pollination Reproductivesuccess .. .. Flg.l.Relauvefrequencyofself-compaublevsself-rncompatible plant speciesshowingnegativefragmentationeffects

self-compatlble specles (Flg. 1), Despite the fact that Table 1 includes aheterogeneous set of studies that differ greatly in habitat type, fragment sizes, surrounding matrix and response variables this

on pollination andreproductivesuccess,P-valuesassociated with a one-tailedFisher's exact test (i.e. are self-incompatible speciesmorelikely to exhibit fragmentationeffectsthan self-compatiblespecies?)are shown.

likely to decline with habitat fragmentation than self-

O

v.

. 888

Aizen, M.A. et al. "-

= =Fragmentation effect.

Table 1. Fragmentationor plant-population size and/or isolation effects on pollination and plant reproductive success.significant negative fragmentation effect; + Species

Family

Acacia aroma

=a significant

positive effect; O=a non-significant effect. FE

Locationa

Habitat type

Fabaceae

Subtropical SA

Dry forest

Tree

SI

Medium-sized to large bees

S

-

-

Aizen & Feinsinger(1994)

Acacia atramentaria

Fabaceae

Subtropical SA

Dry forest

Tree

SI

Bees. beetles

G

O

+

Aizen & Feinsinger(1994)

Acacia brachybotrya

Fabaceae

Temperate Sclerophyllous AU woodland Sbrub

?

Diverse insects

G

?

-

G

O

+

G

O

O

Acacia jurcatispina Acacia praecox Atamisquea emarginata Banksia goodii Brassica kaber Caesa/pinia gilliesi Campanu/a cervicaria Cassia aphylla Centrosema virginianum Cercidium austra/e C/arkia coccinna Dianella

Subtropical SA Subtropical Fabaceae SA Subtropical Capparaceae SA Fabaceae

Proteaceae Brassicacae Fabaceae

Fabaceae Onagraceae

revo/uta

Phormiaceae

Dianthus de/toides

Caryophyllaceae

Embothrium coccineum Enter%bium cyc/ocarpum Eremophi/a g/abra

Eupatorium resinosum Eupatorium perfo/iatum Gentianella germanica lpomopsis aggregata Justicia squarrosa Ligaria cuneifo/ia Mimosa detinens

Temperate NA Subtropical SA

Experimental populations Dry forest Glade Dry forest

Subtropical NA

Mainland vs island

Subtropical SA Temperate NA

Dry forest Disturbed road side

Temperate Sclerophyllous AU

Temperate EU

woodland

Forestand meadow

Fabaceae

Temperate SA Tropical CA

Myopo-

Temperate Sclerophyllous

Asteraceae

Temperate NA

Open wetland

Asteraceae Gentianaceae Po/emoniaceae

Temperate NA Temperate EU Temperate NA

Open wetland Calcareous gtassland Open woodland

Acanthaceae

Subtropical SA

Dry foreSt Dry forest Dry' forest

Proteaceae

raceae

Loranthaceae

AU

Rainforest Dry forest woodland

Tree

SI

Shrub

SI

Butterflies, bees.wasps Bees, wasps Bees,wasps, moths

G

-

SI

Mammals. birds

S

?

-

SI

Bees, fIjes

G

O

O

Hawkmoths

S

-

O

G

?

O

Herb Shrub

SCo%

S

-

O

SC

Large bees

S

-

-

Spears(1987)

SI

Bees, wasps

G

Vine Tree

.. Herb Herb

Herb

SI

Herb

SI

Birds

Wasps,bees, fIjes, moths

SCij%

Hemiparasite

SC¡O%

Sbrub

SCS%

Lythraceae

Boreal EU

Mainland vs island

Subtropical SA

?

Birds. hummingbirds Moths, beetles,bees

Herb

Herb

Lythrum sa/icaria

Portu/acaceae

bees (B)

Herb

Herb

Herb

Portu/aca umbratico/a

Large

Wasps,bees, SI fIjes, moths Flies, SCSS% bees HummingSI birds

Experimental populations

Subtropical NA

butterflies

Bees, SC2Q% butterflies

Tree

Boreal EU

C'!ctaceae

?

SI

Shrub

O

Aizen & Feinsinger(1994)

G

-

-

Groom (1998)

S

?

-

Cunningham (2000)

G

-

-

Jennersten(1988)

S

+

?

G

-

-

Smith-Ramírez& Armesto (in press) Rocha& Aguilar

Bees,fIjes, SC

Tree

Caryophyllaceae

Opuntia stricta

Kunin (1997) Aizen & Feinsinger(1994)

Shrub

Lomiaceae

Subtropical SA

Lamont et al. (1993)

Eisto et al. (2000) Aizen & Feinsinger(1994)

Lychnis viscaria

Cactaceae

Aizen & Feinsinger(1994) Aizen & Feinsinger(1994) Aizen & Feinsinger(1994)

SC

Temperate Disturbed NA deciduousforest Herb

Opuntia quimilo

Cunningham(2000)

Herb

Fabaceae

Onagraceae

SC3%

Reference

Bees, fIjes Large bees(B)

Nepeta cataría

Temperate NA

Specia- FE polli- FE reprod. lizationd nation success

SI

Subtropical SA Subtropical SA

Oenothera macrocarpa

Flower visitorsC

Sbrub

Temperate Sclerophyllous AU woodland Shrub

Boreal Campanu/aceae EU Subtropical Fabaceae SA Fabaceae

Dry forest Dry forest Dry forest

Growth Breeding form systemb

Butterflies

(2001)

S

?

-

Cunningham

G

?

-

Byers (1995)

G

?

O

G

?

..,

S

-

-

Byers (1995) Fischer& Matthies (1998) Heschel& Paige (1995)

S

-

-

Aizen & Feinsinger (1994)

(2000)

Hummingbirds Moths, wasps

S

+

O

G

O

-

Aizen & Feinsinger (1994) Aizen & Feinsinger (1994)

Bees.fIjes, SC64% butterflies

G

-

-

Sih & Baltus (1987)

SC

Bees,butterfIjes, fIjes

G

-

O

Mustaj"árviet al. (2001)

Herb

SI

Bees,fIjes, butterflies

G

-

-

Agren (1996)

Glade

Herb

SI

Hawkmoths

S

-

-

Moody-Weis & Heywood (2001)

forest

Succulent

SI

Medium-sized to large bees

S

O

O

Aizen & Feinsinger (1994)

Mainland vs island Succulent

SI

Bees

G

-

--

Dry forest

Herb

Small bees, SC61% butterflies

G

Spears(1987) Aizen & Feinsinger (1994)

,,~ t.~

- Reproductive

successin fragmented habitats

.

-

889

'"

Table 1, cont.

Species

Family

Location"

Primu/a e/aliar

Primu/aceae

Temperate EU

Forest

Herb

SI

Bees,other insects

G

?

-

Jacquemynet al. (2002)

Prosopis nigra

Fabaceae

Subtropical SA

Dry forest

Tree

SI

Bees,fIjes, wasps

G

-

-

Aizen & Feinsinger (1994)

Temperate

Experimental SI

S

-

-

Steffan-Dewenter&

Herb

Raphanus sativus

Rhipsalis /umbricoides Rutidosis /eptorrhynchoides Salvia pratensis Senna artemisioides Si/ene regia

Cruciferae

EU

Cactaceae

Subtropical SA

Asteraceae

Temperate AU

Habitat type

populations

Dry forest Grassland

Growth Breeding form systemb

Specia- FE polli- FE reprod. lizationd nation success

Solitary

Epiphyte Herb

Fabaceae Caryophy/laceae

Temperate Calcareous EU grassland Shrub Temperate Sclerophyllous AU woodland Shrub Temperate NA Prairie Herb

Sinapis arvensis

Cruciferae

Temperate EU

Experimental populations

Symphonia g/obulifera Spondias mombin

Buttiferae Anacardiaceae

Tropical CA Tropical CA

Rainforest Moist forest

Lamiaceae

Flower visitors'

bees

Reference

Tscharntke

(1999)

SI

Butterflies, bees,wasps

G

-

-

Aizen & Feinsinger (1994)

SI

Beetles,fIjes, moths

G

-

-

Morgan (1999)

S

?

O

Ouborg& van Treuren(1995)

S

?

+

Cunningham(2000)

S

?

-

Menges(1991)

G

-

-

Steffan-Dewenter& Tscharntke(1999)

S

+

+

G

-

-

SC ? SC

Bees Large bees(B) Hummingbirds

Herb

Bees,flies, beetles, SI waps,bugs

Tree

SC

Tree

.SI

Hummingbirds, birds Small diverse insects

Aldrich & Harnrick (1998) Nason& Harnrick (1997)

Ti/landsia BromeSubtropical Dry HummingAizen & Feinsinger ixioides liaceae SA forest Epiphyte SI birds S O O (1994) Tri/lium Temperate Mesic Beetles, Jules& Rathcke ovatum Liliaceae NA forest Herb SI bees,moths G ? O (1999) "AU= Australia, SA= SouthAmerica,NA = North America,CA = Central America, EU = Europe;bSC= self-compatible,SI = self-incompatible.For self-compatible speciesand when information was available,a subscriptindicates%.of either fruit or seedset from flowers wherepollinators were excluded;'B = buzz pollination; dS= pollination specialist,G = pollination generalist.

In our sample (Table 1),50% and 56% of fue trees, 50% and 55% of other woody plan~s(including shrubs, cacti, viDes and one hemiparasite) and 92% and 76% of the herbs showed negative fragmentation effects on pollination (Fisher's Exact Test, P = 0.07) and reproductive success(Fisher's Exact Test, P = 0.20), respec-

Also, reproductive vulnerability in plants is thought to increase with decreasing latitude, becauseof a higher frequency of specialized pollination syndromes (e.g. bat, bird) in the tropics compared with temperate zones (Bawa 1990; Renner 1998). With respect to habitat fragmentation, our data do not support this statement.

tively. This trend is opposite to that expected. Although pollination and reproduction ofherbaceous speciescould be more impaired than in trees due to a more spatially limited pollen flow, this comparison does not support fue hypothesis that compatibility system is an important character in predicting reproductive response to fragmentation.

Are pollination and reproduction success more af -

Although non-significant, we found a trend in the opposite direction with 62% and 52% of fue tropical-subtropical species vs 83% and 71% of fue temperateboreal speciesshowing pollination (two-sided Fisher's Exact Test, P = 0.26) and reproductive (P = 0.23) decline with fragmentation, respectively. However, this trend could be explained by a higher number of studies focusing on herbs in temperate than tropical latitudes (76% vs 10%). The herb growth form shows the highest susceptibility to fragmentation (Table 1). In any event, pollination and reproduction of tropical plants do not seem to be more vulnerable to fragmentation than in

fected in specialist than in generalist plants? Comparisons of fragmentation related changes in pollination and reproductive successbetween specialists and generalists (Table 1) do not support fue hypothesis that specialization in pollination increases fue risk of plant extinction. The proportion of species showing a decline in pollination and reproduction with fragmentation was similar among pollination specialists and generalists (Fig. 2).

temperate plants. One possible explanation for a lack of relationship between specialization and negative fragmentation effects is that pollination specialists reIr only on dependable flower visitors. Waser et al. (1996) demonstrated that fue most successful plants specialize with efficient pollinators that are relatively abundant and exhibit limited variation in time arid space. This would involve~

Habitat fragmentation and pollination specialization

¡;

..

890

c,,",,":

Aizen, M.A. et al. '" "'"

CJ

Specialist Generalist

tion to habitat fragmentation might be attributed to a compensatory association between these two factors. After all, extant plant speciesare fue product of rnillions of years of evolution and should, in general, be relatively adapted to local disturbance and to an ever changing environment. The hypothesis of compensatory effects among reproductive traits was developed by Bond

100 '"'O 80

~ ~ 60 ~ 40

(1994) who showed that, at least in some communities, plants that were severely palIen limited had a low demographic dependenceon seeds,whereas palIen lirnitation was uncommon among plants that reproduce mainly via seed. He also implied fue existence of compensatory effects between pollination specialization and degree of

g It 20 o

.. Polltnatíon

. Reproductlve success

Fi~. 2. Relati~efrequ~ncyofp~llination-~pecialist vs po~linatl0n-general1stspeClesshowmg negatlve fragmentatl0n effects on pollination and reproductive success,P-values . h ' E ' . d Wlt' h a one-tal ' 1e d F IS assoclate er s xact test (l.e, are

reproductive dependence on mutualism (which relates directly th to.,breeding 1d t system) ' but he did not test this idea ' WI

empmca

a a.

.

pollination specialists more likely to exhibit fragmentation effects than pollination generalists?) are shown.

T~e exIstence of co~p~nsatory. e~fec~s between breedmg system and polhnatlon speclahzatlon was not

clearly supported by species listed in Table 1, although the trend was in the expected direction. Whereas we

pollinatorsthattoleratedisturbance and/orarestronger

'

classified50%of theself-compatible species aspollina-

flyers (and are probably less restricted to cross habitat barriers) like many medium and large bees, humrningbirds, bats and hawk moths (Janzen 1971; Stouffer & Bierregaard 1995; Murcia 1996; Aizen & Feinsinger in press), the polliQators usually associated with specialist plants. AIso, plants that utilize pollinators mar have compensatory reproductive traits that allow them to cope with changes in pollinator abundance. In addition to an increasing selfing capacity, these traits could involve an extended life span, profuse vegetative reproduction (Bond 1994) and staggered flowering phenologies (Bronstein & Hossaert-Mc Key 1995). In any event, our results do not support the widespread assumption

tion specialists, this proportiondecreased to 31% among self-incompatible species(one-sided Fisher's Exact test, P = 0.18). Our larger data set provides less support, showing that a similar number of animal speciesvisited fue flowers of both self-compatible and self-incompatible species (including dioecious species). In addition, self-compatible and self-incompatible speciesexhibited a similar frequency distribution in terms of fue number of flower-visiting species (Fig. 3A). This figure also shows that extreme specialization and generalization can be found among both self-compatible and selfincompatible species. However, plants with several mutualist speciesmar still be susceptible if fuese species

that pollination specializationper se increases a plant's vulnerability to fragmentation. Of course, this general statement should not be in conflict with the fact that some specialist plants, such as some island bird pollinated species (Cox & Elmqvist 2000), mar be on the verge of extinction due to fue disruption oftheir pollination mutualism. However, proximate factors other than fragmentation (e.g. introduced diseasesor competitors) are usually involved in the demise of these ¡1lants' pollinators (Renner 1998).

are all taxonomically closely related (Bond 1994). Furthermore, functional specialization in plants mar be better characterized by the number of higher arder animal taxa, particularly at the taxonomic level of"orders" (Johnson & Steiner 2000), Our data set showed that the flowers of>70% ofplant species are visited by only 1-2 orders of animals, with some flowers being visi!ed up to >6 orders of animals. However, there were no significant differences in the m~an number of flower visiting orders or frequency distribution between self-compatible and self-incompatible species (Fig. 3B). It is possible that associations with other factors, such as growth forro or habitat type, could be obscuring a relationship between breeding system and pollination specialization: Although an assessmentof all those potentially confounding factors is beyond the scope of this article, the results of our analyses suggest that the relationship between breeding system and specialization, if present, is not clear cut.

Compensatory effects between breeding system and pollination specialization Can self-incompatible speciesoffset a possible higher vulnerability to fragmentation by being more generalist than self-compatible species?The absenceof independent effects ofbreeding system and pollination specialization on fue responseof plant pollination and reproduc-

¡

, ~

i

.

- Reproductive successin fragmented habitats -

~

-

40 A)

=

891

fuese two characteristics alone, at least in the way they are

sc (9.1~

16, n 841)

/34.039,

P > 0.80

all

SI (8.4~O.9,n=74)

.

d

USU yestlmate

,cannotpre

di h

th

ct ow

erepro

d

.

ucUono f

~

plant specieswill changewith fragmentation. Speciesare characterized by complex suites of integrated traits

g 2G ~ !

(Annbruster et al. 1999), which determine a myriad of interactions and thus fue relevance of different potential causal mechanisms influencing population growth and

~ 30

10

persistence (e.g. variation in seed production,

o

1

2-4

5-10

11-20

persal, seedling recruitment, seed and seedling herbivory; Jules & Rathcke 1999). Therefore, it is unlikely that one

>20

or a few traits and/or ecological processeswill be enough

No. flower-visiting species

to explain why pollination B

fragmentation

60 =

A k I d ts W h kW ' ll ' B d d . C nowe gemen . e t an I lam on an a revlewer for usefu1 suggestions on an ear1y version of the manuscript. Sandra Díaz improved the manuscript with many valuab1e editorial comments: Our research was supported by grants from Consejo Nacional de Investigaciones Científicas y Tecnológicas (CONICET) , Agencia Córdoba Ciencia, Agencia

SI (1,85:t1.11 na130) /34-S.48, P>O~

~

40 >~ 30

6~ 20 10

Nacional de Promoción Científica y Técnica and y Secretaría de Ciencia y Técnica de la Universidad Nacional de Córdoba,

O 1

2

3

4

>4

No, flower-visiting orders.

References

Fig. 3. Histogram of re1ative frequencies of p1ant species ' b f(A)fl ,., d .h ' , assoclate Wlt mcreasmg num ers o ower vIsltmg , specles

and

and reproduction decline with

in many species but not in others.

. -'SC(179f.1tO,na109)

50

-.

seed dis-

, , , (B~ flower-vlsltmg

~rders.

Se,parate

. hls~ogr~ms

for se1f-compaub1e(SC) and self-mcompaub1e plus dloeclous . . I h specles grouped together (SI) are deplcted. n parent eses, means:t. 1 s.e. Results of a G-test comparmg the frequency . . .' dlstnbuuon of SC vs SI specles are shown, ,

Concluding

remarks

, . . The alm of thlS study was to, test common assu~ptlons about how plant reproductlon responds to habItat fragmentation. Particularly, whether or not plant breeding system and pollination specialization influence this response. More studies using comparable protocols, looking at whole assemblages of plants, and measuring the same dependent variables are obviously needed for a

Agren,.J '

1996 P l . . 11' t l . .t t ' .opuauonslze,pomaorlmlalon,ansee th 1f t .b1 h b L h 1"

d

d

'

set m ogy77:

e se -mcompa 1779-1790,

leer

yt

rum sa Icarla.

E co- 1

P 1994 H b' t t fr t u MA & F " A. Izen, .. emsmger,. . a I a agmen a on, 11 ' t' d 1 t od t ' . Ar t. E 1 75. 330gen ma. co ogy. 351. Aizen, M.A. & Feinsinger, P. In press. Bees not to be? Re,

h

po

ma

Ion,

an

p

an

repr

uc

Ion

m

a

caco

dry

..

lores,

t

sponsesof insect po11inatorfaunas and flower po11ination to habitat fragmentation. In: Bradshaw, G .A., Marquet, P. & Mooney, H.A. (eds.) How landscapes change: human disturbance and ecosystem disruptions in the Americas. Springer-Ver1ag,NewYork,NY. Aldrich, P ,R. & Hamrick, J ,L. 1998. Reproductive dominance of pasture trees in a fragmented tropical forest mosaico Science 103-105. Armbruster, W.S., di Sti1io, V .S., Tuxi11,JD., Aores, T.C. &

more "lorma testlng o ese leas. e answer we can . .. , . pro vIde on extant evI~e.n~e IS that no gener~llZatlOnS can be made on susceptlbll1ty to fragmentatlon based on

VelásquezRunk,J.L.1999.Covarianceanddecouplingof floral and vegetative traits in nine Neotropica1 p1ants:a reevaluation of Berg's corre1ation-p1eiadesconceptoAm. J. Bot. 86: 39-55. B K S 1990 PI 11' , " . 1 awa, .. ,ant-po mator mteracUons m troplca rron forests. Annu. Rev. Ecol. Syst. 21: 399-422. Bond, W.J. 1994, Do mutualisms matter? Assessing the im-

compatibility system and pollination specialization. However,ourresults do not mean that fuese characteristics could not be important in determining the demographical decline of particular species, particularly of extreme outbreeders or specialists (see Renner 1998).

pact of po11inatorand disperser disruption on p1antextinction, Phil, Trans, R. SOCo Lond, B 344: 83-90, Bronstein, J.L. & Hossaert-McKey, M. 1995. Hurricane .Andrew and a, F10ri.da fig ~ollin~tion ~utualism: resi1lence of an obhgate mteracUon. Blotroplca 27: 373-381.

N .h

Buchmann,

1

. eIt

er

d

.o

th

.

eymean

...

f th

.d

Th

th tth tr .t h Id tb d a ese al ss ou no euse

. ..

.

1 m assessm~gextln~tlo~ n~k o~ susceptlbll1ty of p~lCU ar plant specIes, WhICh IS slgruficant to conservatlon Bond 1994). However, our results demonstrate

(see that

'

11.

S.L,

&

Nabhan,G,P.

1996.

The forgotten

po lnators. I san 1 dPr ess, C oyel o, CA . Byers, D.L, 1995. Pol1enquantity and qua1ity as explanations for low seed set in sma11populations exemp1ified by Eupatorium(Asteraceae}.Am.J.Bot.82: 1000-1006.

,-

¡

892

Aizen, M.A. et al.

~ ...-

.

j. '-,y

Cox, P.A. & Elmqvist, T. 2000. Pollinator extinction in fue Pacific Islands.Conservo Biol. 1237-1239. Cunningham,S.A. 2000.Effects ofhabitat fragmentationon fue reproductiveecology of tour plant speciesin Mallee woodland.ConservoBiol. 14: 758-768. Eisto,A.J., Kuitunen,M., Lammi, A., Saari,V., Suhonen,J., Syrjasuo,S. & Tikka, P.M. 2000.Populationpersistence and offspring fitness in the rafe bellflower Campanula cervicaria in relation to populationsizeand habitarquality. ConservoBiol. 14: 1413-1421. Feinsinger,P. 1983.Coevolutionandpol1ination.In: Futuyma, DJ.& Slatkin,M.(eds.)Coevolution,pp.282-310.Sinauer, Sunderland,MS. Fischer,M. & Matthies,D. 1998.Effectsofpopulationsizeon performancein the rafe plant Gentianellagermanica.J. Ecol.86: 195-204. Groom,M J. 1998.Allee effectslimit populationviability of an annualplant. Am. Nat. 151:487-496. Herrera, CM. 1988. Variation in mutualisms: fue spatiotemporalmosaicof a pol1inatorassemblage. Biol. J. Linn. Soco35: 95-125. Heschel,M.S. & Paige, K.N. 1995. Inbreedingdepression, environmentalstress,andpopulationsizevariationin scar-

with population size in a fragmented prairie species. ConservoBiol. 5: 158-164. Moody-Weis,J.M.& Heywood,J.S.2001.Pollinationlimitation to reproductivesuccessin theMissouri eveningprimcose,Oenotheramacrocarpa (Onagraceae). Am. J. Bot. 88: 1615-1622. Morgan,J.W.1999.Effectsofpopulationsizeonseedproduction andgerminabilityin anendangered, fragmentedgrassland plantoConserv.Biol. 13: 266-273. Murcia, C. 1995.Edgeeffectsin fragmentedforests:implications for conservation.TrendsEcol. Evol.l0: 58-62. Murcia, C. 1996.Forestfragmentationand fue pollination of neotropicalplants.In: Schelhas,J. & Greemberg,R. (eds.) Forest:patchesin tropicallandscapes,pp. 19-36.Island Press,Washington,DC. Mustajiirvi, K., Siikamiiki, P., Rytkonen, S. & Lammi, A. 200l. Consequences of plant populationsizeand density for plant-pollinatorinteractionsandplant performance.J. Ecol. 89: 80-87. Nason,J.D. & Harnrick,JL. 1997.Reproductiveand genetic consequences offorest fragmentation:two casestudiesof neotropicalcanopytrees.J. Heredity 88: 64-276. . Ouborg, NJ. & van Treuren,R. 1995.Variation in fitness-

let gilia (Ipomopsisaggregata).Conservo Biol. 9: 126-133. Jacquemyn,H., Brys, R. & Hermy, M. 2002. Patch occupancy,populationsizeandreproductivesuccessof a forest herb elatior) in a fragmentedlandscape.Oecologia(Primula (Berl.) 130: 617-625. .

relatedcharactersamongsmall and large populationsof Salviapratensis.J. Ecol. 83: 369-380. Paton,D .C.2000.Disruptionofbird-plant pollination systems in southem Australia. Conservo Biol. 14: 1232-1234. Renner, S.S. 1998. Effects of habitar fragmentation of plant

Janzen,D.H.1971:Euglossinebeesaslong-distance pollinators oftropical plants.Science171:201-203. Jennersten, O. 1988. Pollination in Dianthus deltoides (Caryophyllaceae):effects of habitat fragmentationon visitation and seedser.Conserv.Biol. 2: 359-366. Johnson, SD. & Steiner,K.E. 2000. Generalization versus specializationin plant pollination systems.TrendsEcol. Evol. 15: 140-143. Jules, E.S. & Rathcke,BJ. 1999. Mechanismsof reduced Trillium recruitment along edges of old-growth forest fragments.Conserv.Biol. 13: 784-793. Kattan, G.H. & Alvarez-López,H. 1996. Preservationand managementof biodiversity in fragmentedlandscapesin fue colombian Andes. In: Schelhas,J. & Greenberg,R. (eds.) Forest parches in tropicallandscapes, pp. 3-18. IslandPress,Washington,DC. Kearns,C.A., Inouye, D.W. & Waser, NM. 1998. Endangered mutualisms:fue conservationof plant-pollinator interactions.Annu. Rev.Ecol. Syst.29: 83-112.. Klekowski,E. J. 1988.Mutation developmentalselection,and plantevolution.ColumbiaUniversityPress,NewYork,NY. Kunin, W.E. 1997. Population size and density effects on pollination: pollinator foraging and plant reproductive successin experimentalarraysof Brassicakaber.J. Ecol. 85: 225-234. Lamont,B.B.,Klinkhamer,P.G.L.& Vitkowski,E.T.F. 1993. Populationfragmentationmar reducefertility to zero in Banksia goodii - a demonstrationof fue Al1ee effect. Oecologia94: 446-450. Lloyd, D.G. 1992.Self- and cross-fertilization in plants.11. The selectionof self-fertilization. Int. J. Plant Sci.153: 370-380. Menges,E.S. 1991. Seedgerminationpercentageincreases

pollinator interactionsin fue tropics.In: Newbery,DM., Prins,HE.T. & Brown, ND. (eds.)Dynamicsoftropical communities, pp. 339-360.BlackwellScience,London,UK. Rocha,OJo & Aguilar, G. 2001.Reproductivebiology ofthe dry forestcreeEnterolobiumcyclocarpum(Guanacaste) in

CostaRica:a comparison betweentreesleftin pastures andcreesin continuousforest.Am. J. Bot. 88: 1607-1614. Sih,A. & Baltus,M.S.1987.Patchsize,pollinatorbehaviorand pollinatorlirnitationincatnip.Ecology68: 1679-1690. Smith-Ramírez,C. & Armesto,JJ. In press.Foragingbehaviour of nectar-feedingbirdsvisiting creesin edgesof forest fragmentsandfields in southemChile. Aust. Ecol. Spears,EE. 1987.Islandandmainlandpollination ecologyof CentrosemavirginianulllandOpuntiastricta.J.Ecol.75: 351-362. Spira, T.P. 2001. Plant-pollinatorinteractions:A threatened mutualismwithimplicationsfortheecologyandmanagementof rafe plants.Nat. AreasJ. 21: 78-88. Steffan-Dewenter,1& Tscharntke,T. 1999.Effectsofhabitat isolati9non pollinatorcommunitiesandseedser.Oecologia 121:432-440. Stouffer, P.C. & Bierregaard,R.O. 1995. Effects of forest fragmentationon understoryhummingbirdsin Amazon Brazil. ConservoBiol. 9: 1085-1094. Waser, N.M., Chittka, L., Price, M.V., Williams, NM. & Ollerton, J.1996. Generalizationin pollination systems, andwhy it matters.Ecology77: 1043-1060. Received29 Apri12002; Revisionreceived19 August2002; Accepted21 August 2002. Co-ordinatingEditor: S. Díaz. .