Experimental Induction of Localized Reproduction in a Marine Bryozoan

10 Pages·1993·4.2 MB·English

by  C D Harvell

Reference: Bml Hull 184: 2X6-295. (June,
Experimental Induction of Localized Reproduction
in a Marine Bryozoan
DREW HARVELL AND RICHARD HELLING
C.
Section of Ecology amiSystematic.*. Division oj BiologicalSciences,
Cornell University. Ithaca. New York 14853
Abstract. The control ofreproduction and growth rate locate movingthrougheach zooid. Therateoftranslocate
within colonies of marine invertebrates is often condi- movement through zooids is. in turn, affected by the
tional and can be very localized. We demonstrate exper- strength and proximity ofsinks for that translocate, such
imentally large and localized shifts in the timingand pat- as the growing edge ofthe colony. We propose a simple
tern ofreproduction within colonies ofa temperate bry- source-sink model ofcarbon flow to explain our experi-
ozoan (Membranipora membranacea) in response to mental results. This model would account for the induc-
simulated damage by predators and crowding by conspe- tion of localized reproduction in 1/2 damaged colonies
cifics. In these protandrously hermaphrodite colonies, and the lack oflocalization in 4/8 damaged colonies.
zooidson thedamaged sideofacolony reproduced sooner
than in unmanipulated regions of the same colony. To
Introduction
examine the influence ofthe pattern ofedge damage on
localized reproduction, we damaged the perimeter ofcir- Many organisms do not have fixed life history patterns
cular colonies in two patterns: (1) a continuous half of and are instead plastic in the timing and quantity ofme-
theedgewastrimmed (1/2-Damage) and(2)theedgewas tabolites allocated to growth and reproduction (Cohen.
trimmed in four alternating one-eighth sections (4/8- 1971; Hickman. 1975: Ryland. 1981: King and Rough-
rDeapmraogdeuc)t.ioTnh,ean1d/2t-hdeammoargeeltorceaaltizmeedntfoturri-gegiegrhedthlso-cdaalmiazgeed gHaarrdveenl,l 1a9n8d2:GSrcohslibcehrtgi.ng1,98189:86S:teSatrenasr.ns1a9n89d).KoeIlnlac,ol1o9n8i6a:l
did not. These experiments demonstrate that the config- invertebrates, zooidal responses to biotic and abiotic fac-
uration rather than the total amount ofedge damage af- tors are highly variable because they can be modified,
fects the localization ofreproduction. In parallel experi- sometimes in a very localized region ofa colony, by both
ments, conspecificswereallowedtocrowd halftheperim- intrinsicandextrinsicstimuli (Hughesand Cancino. 1985:
eterofexperimental colonies. This treatment also resulted Harvell and Grosberg, 1988; Harvell. 1991). In fact, the
iznonleocaaldijzaecdeantndtoacacecloenrsapteecdifriecp.roNdoutctioonnlynedaortphaettceornntsacotf oabfilrietsyoutroceasdjausmtotnhgemsohdaupeleosfamroedcuelnetrsalancdhartahcetearlilsotciactsioonf
reproduction change in crowded or damaged colonies, coloniality (Mackie, 1986; Harvell, 1991). For example,
but obstructed colonies also compensate for reduced
the localized induction ofdefensive spines or disruption
growth at an obstructed edge by extending the adjacent
ofa growing edge by damage can induce reproduction in
unobstructed perimeter edge at a greater rate. a marine bryozoan (Harvell and Grosberg. 1988: Harvell
One model to explain the sort oflocal cues governing
and Padilla, 1990; Harvell, 1991. 1992). In hydrozoan
tshoeurocbes-seirnvkedmosdhieflt.s iAn sriempirloadrucmteicohnaannidsmgrioswptrhoproasteedistao (Cnidaria) colonies, a number ofextrinsic stimuli accel-
underlygrowth and reproductiveallocation in plants. We eariaitdeLetnhheoofnfs.et19o5f6r:eBprroadvuecrtmiaonn., s1u9c7h4:asStcerbobwindgi,ng19(8L0o)oamnids
suggest that the balance between growth and onset ofre-
high carbon dioxide concentration (Crowell, 1957; Brav-
production in zooids is determined by the rate oftrans-
erman, 1974). Despite the ubiquity of resource sharing
among modulesofcolonial invertebrates(Mackie, 1986).
Received 30November 1942: accepted 22 February' 1993. there is no theory linking ratesorpatternsoftranslocation
286
LOCALIZED REPRODUCTION IN A MARINE BRYOZOAN 287
there is no theory linking ratesorpatternsoftranslocation ipora are sub-annual, protandrous hermaphrodites; each
and resource allocation to growth and reproduction in zooidproceedsfrom non-reproductivetosperm producing
these organisms. to sperm and oocyte producing to onlyoocyte producing
In other modular organisms such as plants, studies on (Harvell and Grosberg, 1988). As reproduction begins
resource allocation andcarbon budgets have revealed the within the colony, the number of zooids with gametes
importance of an internal balance ofcarbon use in de- and the density ofgametes per zooid is low. This is most
termining ratesoftranslocation and subsequent allocation evident at the onset ofoocyte production, because oocytes
to growth and reproduction (Lang and Thoipe, 1983; can be readily counted. Through time, more zooids de-
Bloom et ai. 1985; Chiarello et at.. 1989; Chapman et velop a greater density ofgametes. Thus increases in the
ai. 1990; Marshall, 1990). Thishasledtothedevelopment densityofreproductive zooids percolonyandoocytesper
of a source-sink model, which forms the paradigm for zooid occur as part ofthe maturation process. In our ex-
carbon (and other metabolite) transport within plants periments, wewill equateagreaterdensity ofreproductive
(Chiarello et ai. 1989; Marshall. 1990). In bryozoans, zooids with a more advanced reproductive state, because
translocation of metabolites from central to peripheral there is no indication from previous studies (Harvell and
zooids has been demonstrated (Best and Thorpe, 1985), Grosberg, 1988; Harvell, 1992) that our experiments
but the relationship between translocation patterns and should affect the quantity of gametes produced. To de-
allocation to growth and reproduction remains unknown. termine whether the timing of reproductive transitions
Our information on resource allocation within colonies varied within different aged regions of undisturbed col-
hasbeen largely limited to patterns ofreproductive timing onies, we monitoredthe reproductivestatesofzooidsfrom
rather than experimental investigations of underlying approximately 40-day-old colonies (n = 11). From pre-
processes. vious studies we knew that 40 days is the approximate
In this paper, we experimentally investigate how en- age when undisturbed colonies begin producing sperm
vironmental factors such as damage and crowding affect (Harvell and Grosberg, 1988; Harvell et at.. 1990). We
the allocation to reproduction and growth in the marine couldagecoloniesbecause we monitoredtheirsettlement
bryozoan, Membranipora membranacea. Weexamine the and growth on lucite panels on which they had naturally
applicability ofa source-sink model ofresource allocation settled, which were suspended underneath the Friday
by analyzing how localized disruptions to a growth sink Harbor (FHL) breakwater (see Harvell and Grosberg,
result inlocalizedshiftsin theonsetofreproduction within 1988, for methods). Colonieson these panel were sampled
colonies. Specifically, we ask how damaging or crowding at three, approximately eight-day intervals in mid July
the colonies' growingedge affects the timingofreproduc- 1987.
tion ofzooids proximal to that edge and how this local The lucite substrateswereeasily manipulated andcould
disruptionaffectsthegrowth rateon adjacent undamaged be brought from their storage locations in the field to the
edges. To examine the role that sink strength may play lab for monitoring. Panels were maintained in a running
in the observed allocation shifts, we vary the magnitude seawater system for the 2-5 h they were in the lab, and
ofdamage by cutting different lengths ofcolony margin during the approximately 20 min sampling were kept
in several experiments. Finally, we use our results to de- submerged incool waterunderthedissectingmicroscope.
velop a source-sink model ofrate ofcarbon flow and re- Reproductive states were monitored under 12X with a
sfuorutrhceerahlylpooctahteisoins-wtietshtiinngaacbooluotnyprtoocaeisdseisnotfherepsrooucrecsesaol-f cWoillodndyisvsaercitaitnigonmiicnrorsepcroopdeuactnidvefibsetratoeptwiacslimghetas.suWrietdhibny-
location within colonial marine invertebrates.
determining the reproductive state of five haphazardly
chosen zooids within each ofthree regions ofthe colony.
Materials and Methods Depending upon the size ofacolony, each ofthese regions
might be populated by greater than a hundred zooids.
Timing ofreproduction Nonetheless, the variance in reproductive state among
To establish the unmanipulated pattern ofonset ofre- the five sampled zooids was low. The sampling was truly
production in undamaged and uncrowded colonies, we haphazardbecausethereproductivestateofazooidcould
mapped the distribution of onset of reproduction for not be determined prior to scrutinizing with a dissecting
zooids within colonies of Membranipora through time. microscopeand zooidsweresampled from all partsofthe
Previous attempts to describe the onset of reproduction particular region. The configuration ofregions sampled
in colonial invertebrates have proceeded with a descrip- from each approximately circular colony was three con-
tive, bulk colony approach (reviewed in Harvell and centric bands ofequal diameter, but decreasing age: an-
Grosberg, 1988; but see Dyrynda, 1981; Wahle, 1983, cestrula region (A)(the founding, morphologically distinct
1984; Brazeau and Lasker, 1990). Colonies ofMembran- "twinned"zooid isattheapproximatecenterofthecolony
288 C. D. HARVELL AND R HELLING
and is called an ancestrula), mid region (M), and edge or Colony areas were determined from tracings of the col-
newgrowth region (E)(Fig. 1). Thewidth ofthese regions onies made at 2-day intervals duringthe 1 1-day duration
was determined by counting the total number of rows of the study. To test the hypothesis that compensatory
containing developed zooids in the colony and dividing growth occurred on the undamaged side ofthe colony in
bythree. Reproductivestateswereeasilyassigned because response to damage, the linear rate ofgrowth was mea-
both sperm and oocytes were visible under microscopic sured on controls and damaged colonies as the distance
magnification through thetransparent frontal membrane from the ancestrula to each opposite edge and the areal
ofthe zooids. Each zooid wasassigned a numerical value growth as the total area added to colonies. These were
for its reproductive state from to 4, and the median of analyzedwith linearregressionofareaandgrowthdistance
thefivevalueswasanalyzed in subsequent statisticalanal- against original colony size. A linear model was used be-
yses. A colony was classified as non-reproductive if the cause both the control and damaged colony data fit this
median statewas or 1 and as reproductive ifthose mea- model well over the range for which we have data.
suresequalledorexceeded 2. Stage 2 wasan unambiguous To determine how localized the stimuli triggering re-
indicator of active reproduction because the sperm are production could be. we also divided 15 moderate-size
mm
actively movingand refract light, showingbrightbirefrin- (approximately 2000 2) colonies into eighths and
gence. damaged alternating one-eighth lengths ofthe perimeter
= (4/8-damage treatment; see Fig. 4 below) and monitored
1 = sonfpoentrh-emrefmprororonudtalulcatemiepvremebsernatnaend visibleonthe underside gargeodwtahndanudndreapmraogdeucdtrieognioonfs.thTehizsoosiedtsofpreoxxpiemrailmetnotsdawma-s
2 = mature spermatozeugmata present (bundles ofac- pnaetrufroarlmlyedononkeeqlupalb-lsaidzeesd,aunndcrroetwudrendedcotloonliiensesgrofofwitnhge
= tively moving and light refractive sperm) breakwater after trimming. The colonies were sampled
3 = oocytes present (and sperm also) once, 16 daysaftertrimming. Forexperimental colonies,
4 only oocytes present. reproductive stateswere recorded forfivezooids 2-4 mm
proximal to each ofthe four damaged and four undam-
Partialdamage experiments
aged sections ofthe colony edge. Eight corresponding re-
New zooids are produced at the outer periphery ofa gionswere sampled on undamaged control colonies from
colony anddevelop from buds to feedingzooidsoversev- the same kelp blades.
eral days. Colonies ofMembranipora are indeterminate
growers and continue to expand at the periphery by pro- Effects ofcrowding by conspecifics
ducing new buds until the colonies begin to senesce in
late summer (Harvell and Grosberg, 1988; Harvell el al., Harvell and Grosberg (1988) showed that the onset of
1990). Although we had previously shown that damage reproduction was accelerated in colonies that were com-
to the margin of a colony accelerates the sexual devel- pletely surrounded by conspecifics. In the current study,
opment of the zooids proximal to the margin (Harvell we designed experiments where colonies were only
and Grosberg, 1988), the degree of localization of this crowded on one side with a single conspecific to assess
effect was unknown and we did not consider how sub- whether the crowding-induced reproduction was localized
sequent growth ofzooids would be affected by damage. like the response to damage. We investigated both the
Could the onset of reproduction be accelerated in one influence ofsize and partial crowdingon timingofrepro-
partofthecolonyin responsetolocalizedcuesand remain duction. We grew 23 pairs ofunequally sized (and aged)
unaffected in anotherpart ofthe colony? It also appeared colonies on lucite panels. This size asymmetry was en-
that growth of a colony was accelerated on a side away gineered by removing all colonies settled on the plate ex-
from damage, so we also tested the hypothesis that com- ceptthoseofparticularsizesand distancesapart. The size
pensatory growth occurred in response to damage. asymmetry wasofno particularimportance tothisstudy,
Workingwith single colonies naturally settled to lucite but it was important for data taken from the same ex-
panels, we trimmed the outer edge buds back approxi- periment and reported in Harvell and Padilla (1990). At
mately one millimeter(equivalent ofonezooid row)with the start ofthe experiment, the small colony ofeach pair
mm
a razor along halfthe perimeter of 14 colonies and mon- was an average size of500 2 and the large colony was
mm
itored them from 13 to 24 July 1987 (1/2 damage treat- an average size of2500 2. Becausegrowth stopsalong
ment). We monitored 13 other undamaged colonies as thecommon borderaftercontact, wedesignated zonesto
controls. At two-day intervals, we determined the repro- sample as in the 1/2-damage treatment. Colonies were
ductive state offive zooids from three regions (A, M, E) sampled on alternate days for 13 days following contact.
on both the damaged and undamaged halfofa colony. Data were pooled into three intervals: <8 days, 8-9, and
LOCALIZED REPRODUCTION IN A MARINE BRYOZOAN 289
100 do not allow us to differentiate timing of reproductive
events in ancestral and mid-colony regions because both
u regionsbecameequally reproductive;zooidsfrom theedge
were still not reproductive (Fig. I ).
Q
Partial damage experiments
In the 1/2-damage experiment, the damaged side pro-
duced both sperm andeggssoonerthan thenon-damaged
ZW side (Fig. 2). Reproduction began on the damaged side
U approximately 5-6 days after trimming (Fig. 2). Unlike
control coloniesand undamaged sidesoftheexperimental
20 I colonies, reproduction was greater at the edge relative to
the ancestral regions on damaged halves. This pattern is
also evident 8-9 days after trimming: more colonies are
reproducing on the damaged side and more colonies are
10 JULY 18 JULY 25 JULY reproducing at the edge than nearer the ancestrula.
DATE Eight to nine days afterdamage (18 July) the effects of
location within the colony are significant, but the effects
Figure 1. Percentage ofcontrol colonies producing spermatocytes of proximity to damage are not (Fig. 2). The location
oroocytesat threedates. Median percentageswerecalculated from five effect is a result ofthe mid and perhaps edge regions be-
z(Ao.oiMd,s hEap=haAznacredsltyrals.amMpilde.dafnrdomEdegaechreogfiotnhsr)e.eBroetghiownistihnin1-c1ocloolnoynileos- ginning to reproduce, but the center not (Fig. 2).
cation(X2= 10.49,P=0.005)anddate(\- = 7.91.P=0.02)significantly Three days later (11-12 days after damage) the effects
affectthe frequency ofreproductivecolonies. of proximity to damage are pronounced (Fig. 2). The
within-colony location term is no longer significant al-
though inspection ofFigure 2 does reveal a tendency for
10-13 daysaftercontact. Five zooidswereagain sampled reproduction to vary with location within the colony.
from each region, and the median of those values ana-
lyzed. Because theshapes ofcoloniesbecame sodistorted
in the crowding treatment due to compensatory growth 80
away from the interaction, it was impossible to consis-
tently sample the same three regions as on the controls
(A, M, E). We therefore divided the crowded colonies
into regions oftwo (M, E) instead ofthree equal radii for
sampling.
Results
Timing ofreprodiiction
Under normal, good-growth, field conditions, colonies
begin reproduction as males at an age ofapproximately
mm
40-60 days and would be approximately 2500 2 in ft M5-6E DAYftSM E I A M8-9E DAY6S M E I A11M-12E DAYASM E
area (Harvell. 1992). The pattern ofreproduction in un-
NUMBER OF DAYS POST MANIPULATION
manipulated colonies is for older zooids near the ances-
trulaand in the mid regions ofcolonies to reproduce first Figure 2. l/2-damage Experiment: median percentage ofcolonies
(Fig. 1). Most zooids are producing spermatocytes and pforrodduacmianggedooc(yotpeesnobrarssp)earmnadtoucnydtaems.agReedpr(osdhuacdteidvbearsst)atheaslwveerseoftacboulloantieeds
oocytes by late July. and three regions within each colony (A, M, E). Sample size at top of
The significant results ofthe log-linear test reflect the eachbar. At 5-6daysthefrequencyofreproductivecoloniesisidentical
transition in ancestral and mid regions ofcolonies from fordamaged and control colonies. At 8-9 days, within-colony location
a25noJnu-lryep(Friogd.uc1)t.ivBeyt2o5reJpurlyo,dutchteivpeersctaetnetabgeetwoefenanc1e8starnadl appfrrfooexxctiismmiittthyyetftoroeddqauamemnaacggyeeoafdtofreeecsptrsnotodhtuec(tfXirv:eeq=uceo2nl.oc5ny9i.eosfP(rXe-p:r0=o.d17u.1c)1t.2i,AvtePc<1o1l0-o.1n20i3e)dsaby(usXt,2
and mid-colony regions that are reproductive has in- = 9.05, P = 0.003), but within colony location does not (X2 = 2.22, P
creased from to 25%, respectively, to 100%. These data = 0.33).
290 C. D. HARVELL AND R HELLING
4/8- DAMAGE TREATMENT testcomparingthe4/8-damage andtheircontrolsdetected
no significant differences in the within colony variation,
but a highly significant treatment effect, confirming a
generalized increase in reproductive activity in the 4/8
damage colonies (Fig. 3).
Effects ofcrowding by conspecifics
In pairedcoloniessharingacrowdededge, the response
ofcolonieswasanalogoustothe 1/2-damage experiment.
Zooidson theobstructed sideofacolonyshowed a higher
median percent reproduction than zooidsfrom the unob-
structedsideofthecolonyat8-9days(Fig.4). Forcolonies
incontact forlessthan eightdays, therewerenosignificant
differences in reproduction (Fig. 4). At 8-9days, colonies
crowded with a single conspecific began to reproduce lo-
cally near the contact point. In the log-linear test offre-
quencies at 8-9 days, the within colony location and the
proximity terms were significant (Fig. 4).
UNDAMAGED CONTROL The pattern was different at 10-13 days after contact.
The frequency of reproduction was still higher on the
contact sidethan thenon-contactside. Thewithin colony
localization term was clearly significant, with the edge
region in both treatments showing the highest reproduc-
tion (Fig. 4).
Compensatorygrowth
Patterns and rates ofgrowth on untrimmed halves of
We
colonies were also affected by the manipulation.
Figure3. 4/8-damage Experiment: median percentageofcolonyre- H 60.
g1i6ondsaypsroadfutcerindgaomoacgyet.esNourmsbpeerrmsataorceyttheesimne4d/i8a-ntripmeracnedntcaognetrooflccoolloonniieess U3oa 16
producingspermatocytesoroocytes.Thefrequencyofreproductivecol- a-t
onies varies in 4/8 damaged colonies relative to control colonies (X2 _
= 28.41, P = 0.00): proximity to damage does not significantly affect
reproductivetiming(X2 = 0.03. P = 0.86). U
aet
Proximitytodamage isclearly astrongeffect, the median
percent ofreproductive colonies isdouble adjacent tothe ME ME
damaged edge in comparison to the undamaged sides of <8 DAYS 10-13 DAYS
the colony (Fig. 2). DAYS FROM FIRST CONTACT
The pattern of reproduction in the 4/8-damaged col-
onies was different from the 1/2-damaged colonies. All Figure 4. Crowding Experiment: median percentage of colonies
the colonies were sampled 16 days after the edge was producingoocytesorspermatocytesonobstructedandunobstructedsides.
Open bars are obstructed sides, closed bars are unobstructed sides of
trimmed. Although a higher proportion oftrimmed col- colonies.Samplesizeattopofeachbar.At<8dayspostcontact,neither
onies than untrimmed were reproductive, there was no locationwithincolony(X2 =0.00.P=0.95)norproximitytoconspecifics
regionalization ofreproduction within the damaged col- (X2 = 0.35.P= 0.55)affected frequencyofreproductivecolonies. At8,
oangieedsc(oFliogn.i3e)s.,Oruerpreoxdpueccttiaotniownowualsdtbheatl,ocliakleiztehdea1n/d2-pdraomx-- a9nddaypsropxoismtitcyonttoaccto,nsbpoetchifliocca(tXi2on=w4i.t9h9,inPc=ol0o.n0y2)(Xs2ig=nif4i.c9a9n,tlPya=ff0e.c0t2e)d
imal to the trimmed edges. Instead, reproduction was tlhoecaftrioenquweintchyinocforleopnryod(uXc2ti=v4e.c4o9l.oPnie=s.0.A0t3)1a0n-d13prdoaxyismiptosyttococnotnascpte,cibfoitchs
uniformly accelerated in all zooids sampled, irrespective (X2 = 7.14.P= 0.007)significantlyaffectedthefrequencyofreproductive
of their proximity to the damaged edge. The log-linear colonies.
LOCALIZED REPRODUCTION IN A MARINE BRVOZOAN 291
COMPENSATORY GROWTH We also compared the area added to the l/2-damage
and control colonies over the same time interval. The
total area added to damaged colonies was less than that
added to undamaged colonies, asindicated by differences
in slope ofthe initial area against area added plot (Fig.
6b). The ANCOVA showed significant effects ofthe co-
variate and slope differences between the two treatments
(Table Ib). Therewas noeffectofdamageontheintercept
intheareaanalysis. Coloniesthusresponded totrimming
by not only accelerating the onset ofreproduction in the
damaged half, but also by accelerating growth in a direc-
tion away from the damage.
Discussion
Despite a widespread recognition that the population
dynamics and biology ofmodular organisms differ from
thoseofunitary organisms, on which mostecological the-
ory is based (Harper. 1981; Jackson el ai. 1985; Harper
el a/.. 1986), many fundamental aspects oftheir biology
are little studied. These aspects include the physiological
integration ofcolonial invertebrates, the nature ofthe in-
CONTROLS (N = 13) 1/2-DAMAGE (N = 15) teraction between zooid and colony level processes, and
particularly inter-zooid allocation phenomena. Wahle
Figure5. Diagrammaticdepictionoftheappearanceofcompensator,'
growthinthe l/2-damagetreatment.Theshadedareaontheinitialhalt-
damage colonies indicates where the edge was trimmed. The compen-
satory growth of the half-damaged colonies was detected as a greater DSEMI-DAMAGE
edgeextension rate relativetoan undamaged control colony.
monitored growth in the damage experiments and found
that colonies responded to localized damage by directed,
compensatory growth. Figure 5 showsa representation of
compensatorygrowthdennedasan increased area-specific
edge extension rate. The increase in edge extension was
detected on the undamaged side of1/2-damaged colonies
relative to control colonies (Fig. 6a). When growth was 200 400 600 800INITI1A0L0A0REA1200 MOO 1600 1800 2000
disrupted by trimming on one side, intact sides of the QSEMIDAMAGE
colony responded with an elevated edge extension rate.
Thesize-specific rateofextension ofa normal colonywas
less than the rate ofextension ofthe untrimmed halfof
an experimental colony, as indicated by the difference in
intercepts ofthe regressions ofedge extension on initial
area (Fig. 6a). This indicates that the undamaged edge
was growing faster on damaged than on undamaged col-
onies. The analysis ofcovariance [ANCOVA] revealed a
significant effectofthecovariate(initial area) in the model 800 1000 1200 1400 1600 1800 2000
and no significant difference in the slopes (included in INITIALAREA
the initial area*damage term) (Table la). The damage Figure 6. (Top) Compensatory growth measured as an increase in
treatment significantly affected theintercept when the in- thelinearextension rate in l/2-damagedcolonies. Linearregression for
significant slopeterm wasdropped from the model. Thus ccoonltornoilescoisloYnie=s0is.Y000=908.X00+0I\X.5+606(.j9291=(.r691=)..7(8B3o)tatnodm)foCroemxppeenrismaetnotrayl
irrespective of colony size, the rate ofedge extension is growthmeasuredasareaaddedto 1/2-damagedandundamagedcolonies.
greater forpreviously damaged than control, undamaged Linearequation forcontrolcoloniesisY =0.336X +44.289(r = .972)
colonies. and forexperimental coloniesisY = 0.135X + 33.089 (r = .978).
292 C. D. HARVELL AND R. HELLING
Table I
Analysisofcovariance[ANCO\'.!] <m cnmpensatorygrowth
SOURCE DF TYPE III SS MS
EDGE EXTENSION
LOCALIZED REPRODUCTION IN A MARINE BRVOZOAN 293
CONTROL
tracer studies (Best and Thorpe, 1985; Miles, Harvell. A.
Griggs, and Eisner, unpub.). Furthermore, even colonies
fed near the growing edge only translocate to the nearest
edge, confirming that transport is unidirectional (Miles,
Harvell, Griggs, and Eisner, unpub.).
The simplest model of translocation assumes equal
transport through all pores of a zooid, with rates deter-
mined by sink strength. Because axial pore plates are
slightly different than lateral plates morphologically, it is DAMAGE
B. 1/2
not unreasonable to hypothesize higher transport axially EDGE
than laterally. Both because of a bias to axial transport
and because the growing margin ofa colony is a strong
carbon sink, transport rates should be greatest in a distal
direction. In Figure 7 ahypothetical model ofthedirection
and quantity ofcarbon flow through zooids for each of
the sink disruption treatments is shown. Although trans-
port is normally polarized in a proximo-distal direction,
this polarity can presumably be reversed to support re-
growth in proximal regions ofthe colony when these are
damaged (Jackson and Palumbi. 1979; Harvell, 1984). DAMAGE
C. 4/8
When the edge sink is disrupted, the polarity oftranslo- EDGE f UNCUT
EDGE
cation should change and the flow of carbon either be
redirected to the next most active sink or zooids at the I MED MED _1_
edge should reproduce with the surplus carbon. In the Si.*"*-P
case ofthe 1/2-damage treatment, the next sink is on the
other side of the colony and so the response might be
reverse translocation. Because the sink is so distant, the
translocation isexpected to beweak and zooids keep most
ofthe carbon they take in (Fig. 7B). In the case of the Figure7. Thesource-sink model ofcarbonallocation incheilostome
4/8-damage treatment, the next sink is adjacent to the bryozoans.Thehypothesizedallocationoftranslocateisdepictedtorthe
short lengthofdisruptededgeandisstill incloseproximity three experimental treatments described in this paper. A: Control, B:
to zooids via the lateral pores (Fig. 7C). Thus we hypoth- 1/2-damage(onehalftheperimeterofacolonyremovedinacontinuous
esize that the zooids in the 4/8-damage experiment re- section),C: 4/8-damage(one halftheperimeterofacolony removed in
four alternating sections separated by four sections ofundamaged pe-
mained nonreproductive because active sinks close by nmeter). The thicknessofthearrows isproportional tothequantity of
continued to use metabolites. Such a mechanism would carbon transported acrosszooidal boundaries. The labels, Lo, Med, Hi
produce the results we observed: (1) the onset of repro- designate the relative amount oftranslocate retained by a zooid. Thus
duction in parts ofthe half-damage colonies, (2) the lack forcontrol colonies, only a low proportion ofthecarbon is retained by
ofregionalization in the 4/8-damage experiment, and (3) atrzeoaotimde.ntI,nacnodntrtahsetr,ewihsennotchleosseinrkepisladciesmreunpttedsianks,inthtehecahrablofn-diasmangoet
thecompensatorygrowth observedin the 1/2-damageex- translocatedandzooidsretainahighproportion. Thehigherproportion
periment. Although colonies in the 4/8-damage treatment ofretained carbon could be used togrowgametes, thusaccountingfor
showed no regionalization in timing reproduction, there localized reproduction in zooidsadjacenttodisruptedsinks. Inthe4/8-
was a slight increase in reproduction throughout the col- damage, theoriginal sink isdisrupted butan adjacentsinkstillstrongly
affectszooid translocation.
ony relative to the controls. This is consistent with our
model; removinghalftheperimetersloweddowndiffusion
ofmetabolitesthroughout theentirecolony, but no region
was far enough from a sink to show localization. rate was not sufficient to allow the damaged colonies to
Wehavedemonstrated notonlylocalizedaccelerations add thesame areaasthe undamaged coloniesofthesame
of reproduction within colonies, but also an associated size. A localized increase in edge extension adjacent to
and localized acceleration in edge extension rate in non- an obstructededge onceagain suggestsanextremely plas-
damaged regionsoftrimmedcolonies. Toourknowledge, tic, colony-wide source-sink allocation budget. We hy-
suchcompensatorygrowth(an increasedarea-specificedge pothesize the following mechanism to account for the
extension rate) has notbeen shown previously forcolonial shift. Becausetheedgeofthecolonyisnormallysubsidized
invertebrates. In this case, the increased edge extension by proximal zooids and thus represents a sink for metab-
294 C. D. HARVELL AND R. HELLING
elites, removingtheedgedisruptsthe flowofmetabolites. Chiarello.N.R.,11.A.Mooney,andk.Williams. 1989. Growth,carbon
Some ofthe carbon is used locally by zooids for repro- allocation and cost ofplant tissues. In Plant Physiological Ecology:
duction andsomeofthecarbon isredirectedtothe nearest FHi.elAd.MMeotohnoedys aannddIPn.sWtr.umReunntdaetli,one,dsR..CWh.apPmeaarny.anJ.dRH.alElh.leringer.
sink the undamaged edge. In thissituation, the undam- Cohen. D. 1971. Maximizing final yield when growth is limited by
aged edge adjacent to damage can grow more rapidly, timeorby limiting resources.J. Theoret. Biol 33: 299-307.
because it is supplemented by more zooids than normal. CroMell, S. 1957 Differential responses ofgrowth zones to nutritive
This also suggests that normal colony growth rates are level, age. and temperature in the colonial hydroid Campanularia.
limited bymetaboliteavailability and that ifmore internal DyryJndEax,p.P.7.Eoo.l.J.13149:816.3-9A0.preliminary study ofpolypide generation-
zooids could be added to subsidize the edge then the col- degeneration in marine cheilostome bryozoa. In Recent and Fossil
ony would extend across the substrate at a greater rate. Bryozoa.G. P. LarwoodandC. Nielsen,eds. OlsenandOlsen.Fre-
An important consequence ofmodularity may be the denburg. Denmark. P. 79.
ability to continually re-allocate the priority ofresource Harper.J. I.. 1981. Theconceptofpopulationinmodularorganisms.
transport amongthe unitsofacolonyastheenvironment Pp. 53-77 m Theoretical Ecology: Principles and Applications,
changes. The existence of thresholds within zooids that R. M. May, ed. SinauerPress. Sunderland, MA.
determine major shifts in physiology from growth to re- Harpfeorr,mJo.fLm.,odBu.lRa.rRoorsgeann,isamnsd.JP.hiWl.hiTtrea,nse.dsR.o1y9a8l6.Sac.ThSeer.grBow3t1h3.and
production, permit zooid specialization, depending upon Harvell, C. 1). 1984. Why nudibranchsare partial predators: intraco-
their location within the colony. The nutrient flow pat- lonial variation in bryozoan palatability. Ecology65: 716-724.
ternswe have proposed remain hypothetical, butare test- Harvell, C. D. 1991. Coloniality and inducible polymorphism Am
able usingradioisotopestotrace the fateofnewlyacquired Nat. 138: 1-14.
carbon (Best and Thorpe, 1985: Miles, Harvell, Griggs Harvell,C.D. 1992. Inducibledefensesandallocationshiftsinamarine
and Eisner, unpub.). We propose that these allocation Harvberlylo,zCo.anD..,EacnodloR.gyK.73G:ro1s5b6e7r-g1.517968.8. Thetimingofsexualmaturity
capabilitiesaregeneral forcheilostome bryozoansand, in in dona! animals. Ecology69: 1855-1864.
some form, for all colonial invertebrates. Further exper- Harvell, C. D., H. Caswell, and P. Simpson. 1990. Density effects in
imental work is needed to test models of energy flow acolonial monoculture:experimentalstudieswithamarinebryozoan
through colonies in relation to taxonomic identity and (Membranipora membranacea L.). Oecologia82: 227-237.
levels ofphysiological integration. Harvell, C. D., and D. K. Padilla. 1990. Inducible morphology, het-
erochrony,andsizehierarchiesinacolonial monoculture. Proc. Nail.
Acad. Sci 87: 508-512.
Acknowledgments Hickman, C. J. 1975. Environmental unpredictability and plastic al-
locationstrategiesintheannualPolygonumcascadensc(Polygonacea).
This work was supported by NSF OCE-8796230 and J. Ecol 63:684-701.
OCE-8817498. Assistance with the field sampling was Hughes, R. N., and J. M. Cancino. 1985. An ecological overview of
parnodvwiidtehdsbtyatDis.tiKc.alPaandailllyas,iswibtyhMd.atGaebanearl.ysTihsebymaCn.uGsrcirgigpst, oecd/ls'o.CnliYonanglaielnOUmnreigtvaaenzriossaim.tsyP.pP.Jr.e1sBs5.,3-CN.1e8Jw6aciHknasPvoonep.nu.lLa.CtTWi.o.nBBuisosl,ogayndanRd.EEv.olCuotoiko.n
was improved by comments from D. Shapiro, M. Geber. Jackson,J. B.C., L. \V. Buss,and R. E.Cook,eds. 1985. Population
M. Bertness. BiologyandEvolution ojClonalOrganisms. Yale University Press.
New Haven,CT.
Jackson, J. B.C., and S. R. Palumbi. 1979. Regeneration and partial
Literature Cited
predationincrypticcoralreefenvironments:preliminary experiments
Besta,nMd.tAh.e,caonldonJi.alP.tTrhaonrsppoer.t1o98f5.metaAbuotloirtaedsioignrathpehimcasrtiundeyobfryfeoezdoianng ColnasupdoengLeesvianadndecNt.oprBoocutrsg.-EPspn.au3l0t3,-3e0ds9.iCneBnitorleogNiactidoensaSlpodnegilaarRese.-
Membranipora membranacea. Mar. Biol. 84: 295-300. chercheScientifique, Pans, France.
Bloom, A.J., F. S. Chapin, and H. A. Mooney. 1985. Resource limi- King, D., and J. Roughgarden. 1982. Multiple switches between veg-
tation in plants an economic analogy. Ann. Rev Ecol. Sysl. 16: etativeandreproductivegrowihinannualplants. Theoret.Pop.Biol
21: 194-204.
363-392.
Bobin, G. 1977. Interzooecial communicationsand the funicularsys- Lang, A., and M. R. Thorpe. 1983. Analysing partitioning in plants.
tem. Pp. 307-333 in Biology ofBryoioans. R. M. Woollacott and Plant Cell. Environ. 6: 267-274.
R. L. Zimmer, eds. Academic Press, New York. I iii.mil.(. 1977. Thebryozoannervoussystem.Pp.377-410inBiology
Braverman. M. 1974. The cellular basis ofmorphogenesis and mor- olBrvo:oans, R. M. Woollacott and R. L. Zimmer. eds. Academic
phostasisin hydroids. Oceanogr Mar Bid/ 12: 129-221. Press. New York.
Brazeau, D. A.,and H. R. 1-asker. 1990. Sexual reproduction andex- Loomis, W. F., and H. M. Lenhoff. 1956. Growth and sexual dif-
ternal broodingbytheCaribbean gorgonian. Bnarcumasbeslinum. ferentiation of Hydra in mass culture. J Exp. Zoo/. 132: 555-
Mar Bioi 104: 465-474. 568.
Chapman,D.F.,M.J.Robson,and R.\V.Snaydon. 1990. Short-term Mackie,G.O. 1986. Fromaggregatestointegrates:physiologicalaspects
effectsofmanipulatingthesource-sinkratioofwhiteclover(Trifolium ofmodularity incolonial animals. Pp. 175-196inTheGrowth and
repens L.) plantson export ofcarbon from and morphology of,de- FormofModularOrganisms.J.L. Harper.B.R.Rosen,andJ.White,
velopingleaves. Physio/ flam80: 262-266. eds. The Roval Societv. London.
LOCALIZED REPRODUCTION IN A MARINE BRYOZOAN 295
Marshall,C. 1990. Source-sink relationsofinterconnected ramets. In Slehbing,A. R.I). 1980. Increasesingonozooid frequencyasanadap-
ClonalGrowth inPlants. RegulationandFunction, J. van Groenen- tive response to stress in Campanularia HCMIUMI Pp. 27-32 in De-
daal and H. de Kroon. eds. Academic Publishing. The Hague. The velopmentaln/iilCellular Biology <>/ Coelenlcratc*. P. Tardent and
Netherlands. R. Tardent, (eds.). Elsevier. New York.
KM.Hid. J. S. 1981. Colonies, growth and reproduction. Pp. 221-226 \\ahlc, C. M. 1983. Regeneration of injuries amongJamaican gor-
in Recent amiFossil Br\o:oa. G. P. Larwood and C. Nielsen, eds. gonians:therolesofcolonyphysiologyandenvironment. Bin/. Bull.
Olsen andOlsen, Fredensburg. 165: 778-790.
Schlichting,C.D.1986. Theevolutionofphenotypicplasticityinplants. \\ahle, C. M. 198-1. Scaling problems in the ecology ofcolonial in-
Ann Rev Ecol Sy-,1 17: 667-693. vertebrates: injury, mortality, and colony integration in gorgonian
Stearns,S.C. 1989. Theevolutionary significanceofphenotypicplas- octocorals. Ph.D. Dissertation,Johns HopkinsUniversity, Baltimore,
ticity. BiiHcience39: 436-446. MD.
Stearns. S. C., and J. C. Koella. 1986. The evolution ofphenotypic Watson, M. A., and B. B. Caspar. 198-1. Morphogenetic constraints
plasticity in life history traits: predictionsofreaction norms forage on patterns ofcarbon distribution in plants. Ann. Rev. Ecol. Syst.
andsizeat maturity. Evolution40: 893-913. 15: 233-258.