ECOLOGY OF THE OBLIGATE SPONGE-DWELLING BRITTLESTAR Ophiothrix lineata
Timothy P. Henkel
A Dissertation Submitted to the University of North Carolina Wilmington in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
Department of Biology and Marine Biology
University of North Carolina Wilmington
2008
Approved by Advisory Committee
Martin Posey ______John Bruno ______
Fred Scharf Ami Wilbur ______
______Joseph R. Pawlik Chair
Accepted by
______Dean, Graduate School
i
TABLE OF CONTENTS
ABSTRACT ...... v
ACKNOWLDEGEMENTS ...... viii
LIST OF TABLES ...... ix
LIST OF FIGURES ...... xi
CHAPTER 1: THE ASSOCIATION OF Ophiothrix lineata AND Callyspongia vaginalis: IS
THE BRITTLESTAR A PARASITE ON SPONGE LARVAE? ...... 1
ABSTRACT ...... 2
INTRODUCTION ...... 3
MATERIALS AND METHODS ...... 5
Predation of sponge larvae by Ophiothrix lineata ...... 5
Effect of Ophiothrix lineata on Callyspongia vaginalis ...... 6
Statistical analyses ...... 9
RESULTS ...... 10
DISUSSION ...... 11
LITERATURE CITED ...... 16
CHAPTER 2: HOST SPECIALIZATION AND LIMITATION OF AN OBLIGATE SPONGE-
DWELLING BRITTLESTAR ...... 25
ABSTRACT ...... 26
INTRODUCTION ...... 27
MATERIALS AND METHODS ...... 30
Host choice experiments ...... 30
Effect of habitat size on growth of Ophiothrix lineata ...... 32
ii
Growth of Ophiothrix lineata living on multiple sponge hosts ...... 34
Sex and size distribution of Ophiothrix lineata ...... 35
RESULTS ...... 36
DISUSSION ...... 37
Host specificity of Ophiothrix lineata ...... 37
Host Limitations and the distribution of Ophiothrix lineata ...... 43
LITERATURE CITED ...... 45
CHAPTER 3: LIFE HISTORY TRAITS OF THE SPONGE-DWELLING BRITTLESTAR
Ophiothrix lineata ...... 57
ABSTRACT ...... 58
INTRODUCTION ...... 59
MATERIALS AND METHODS ...... 61
Larval development ...... 61
Morphometrics and size distribution ...... 61
Growth model ...... 62
RESULTS ...... 64
Larval development ...... 64
Morphometrics and size distribution ...... 65
Growth model ...... 66
DISCUSSION ...... 66
Development of Ophiothrix lineata ...... 66
Implications of direct development in Ophiothrix lineata ...... 69
Growth of Ophiothrix lineata ...... 70
iii
LITERATURE CITED ...... 73
CHAPTER 4: THE EFFECT OF AN OBLIGATE SPONGE-DWELLING BRITTLESTAR ON
THE SPONGE-ASSOCIATED COMMUNITY ...... 83
ABSTRACT ...... 84
INTRODUCTION ...... 85
MATERIALS AND METHODS ...... 86
Surveys of established community ...... 86
Immigration of sponge-associated community ...... 87
Statistical analyses ...... 88
RESULTS ...... 88
Surveys of established community ...... 88
Immigration of sponge-associated community ...... 89
DISCUSSION ...... 90
LITERATURE CITED ...... 93
iv
ABSTRACT
Host specificity has long been recognized as an important means of speciation in terrestrial plant-herbivore and host-parasite interactions; however, the role of host specificity in the marine environment is less understood. On coral reefs off the Florida Keys, the sponge- dwelling brittlestar Ophiothrix lineata lives almost exclusively in the tube sponge Callyspongia vaginalis. I examined chemotactic recognition by O. lineata to assess sponge host preferences using a y-tube assay chamber. Relative to seawater controls, O. lineata preferentially selected seawater conditioned by the preferred host sponge C. vaginalis and showed no preference for seawater conditioned by the infrequent host Niphates digitalis or non-host Aplysina archeri.
When offered seawater conditioned by C. vaginalis and N. digitalis, O. lineata chose C. vaginalis 78% of the time. Growth of O. lineata living in the three sponge hosts was measured in the field. Growth of O. lineata was greatest when living in C. vaginalis, though growth was similar to O. lineata living in N. digitalis. Unlike A. archeri, both N. digitalis and C. vaginalis brood larvae year round. Sponge larvae may provide an additional food source to O. lineata, which deposit-feeds on the outer surface of the sponge.
Predation of sponge larvae by O. lineata was first assessed in lab feeding assays, and O. lineata readily consumed larvae from both C. vaginalis and N. digitalis. In the field, O. lineata grew faster when living in both species of brooding sponges than when living in non-brooding individuals. Larval predation by O. lineata in the field was further assessed by comparing larval output from single tubes of C. vaginalis with and without associated O. lineata. Variation in larval output was high among tubes of C. vaginalis. Sponge tubes without O. lineata released more larvae and were more variable in larval output relative to sponge tubes with O. lineata. In addition, there were significantly fewer larvae relative to the number of brood chambers in C.
v
vaginalis with O. lineata present. Predation of sponge larvae by O. lineata may be reduced due to the high variability in timing of larval release, random distribution of brood chambers throughout the sponge, and even distribution of O. lineata across brooding and non-brooding C. vaginalis.
Host specificity in O. lineata may also be a function of sponge morphology, as the multi- tubed shape of C. vaginalis can provide habitat to multiple O. lineata. Like most echinoderms,
O. lineata releases gametes into the water column and fertilization success may be correlated to density of conspecifics. The sex distribution of O. lineata within C. vaginalis was surveyed. Of the 35 sponges surveyed, 74% had at least one male and female O. lineata, with 37% of sponges having a greater proportion of males, both of which may increase brittlestar fertilization success.
The evolution of the obligate association of O. lineata with C. vaginalis likely resulted as a consequence of some combination of host abundance, enhanced food availability, and greater probability of mating success in the multi-tubed sponge.
Development and growth of O. lineata was also examined. In the lab, O. lineata spawned and fertilized eggs were monitored over time. Fertilized eggs were 330 μm in diameter and embryos within the fertilization membrane averaged 230 μm. Within 6-8.5 days, O. lineata underwent complete development within an adhesive fertilization membrane, after which juveniles ~ 150 μm disk diameter hatched. This is the first report of non-pelagic direct development in the genus Ophiothrix, but morphological similarities suggest that this developmental mode may be common to other species. Growth of O. lineata was measured using a range of initial disk diameters (5-11 mm), and growth models were used to calculate size-at-age estimates. Average specific growth rate was 3.2 mm yr-1± 4.9 SD, and decreased
significantly with increasing disk diameter. Growth rates were highest for brittlestars 5-8.5 mm
vi disk diameter, coinciding with average size of O. lineata. Growth models yielded estimates of 1 year of growth as 5-8.3 mm disk diameter.
The effect of O. lineata on both the host sponge and sponge-associated community was also assessed through a series of field experiments. In manipulations of the community living in
C. vaginalis, there was no impact of O. lineata on immigration of other species. Growth and reproduction of C. vaginalis living with and without associated O. lineata was examined in the field. After 8 months, there was no difference in the number of brood chambers in sponge tissue or in overall sponge growth of C. vaginalis with or without O. lineata. Since O. lineata may consume larvae of C. vaginalis, the association between O. lineata and C. vaginalis varies from a simple commensalism to a larval parasitism depending on the reproductive status of the sponge.
vii
ACKNOWLDGEMENTS
This work was made possible through the support and guidance of Dr. Joseph R. Pawlik.
His commitment to the scientific process has enhanced my training as a scientist and an educator.
In addition, I benefited greatly from the insights and thoughts of my committee, Drs. Martin
Posey, Fred Scharf, Ami Wilbur, and John Bruno. I am also appreciative to all the faculty, graduate students, and staff in the Biology and Marine Biology department for conversations that
sparked new ideas and tested old ones. This research was funded by grants to Joseph R. Pawlik from the National Undersea Research Program at UNCW (NOAA NA96RU-0260) and from the
National Science Foundation, Biological Oceanography Program (OCE-0095724, 0550468).
As in any field intensive project, this work would not have been possible without a great
team of field assistants. I am extremely grateful to Steve McMurray for his tireless pursuit of the
sponge-dwelling brittlestar. I also thank Thor Dunmire, Wilson Freshwater, Adam Jones, Doug
Kesling, Wai Leong, Tse Lynn Loh, Greg McFall, Otto Rutten, Jay Stryon, Susanna Lopez-
Legentil, Sven Rhode, Kyle Walters and the entire staff at the National Undersea Research
Center in Key Largo Florida for assistance in the field. Research on the life history of
Ophiothrix lineata was done in collaboration with Vince Richards of Nova Southeastern
University, who documented the development of O. lineata.
Finally, I would not have returned to this work without the support and faith of my
family. I could not have continued it without my friends nor have enjoyed it as much without
Milo and Pepper. And I would never have finished it without the love, patience and support of
Sharon.
viii
LIST OF TABLES
Table Page
1. Timing of long and short term experiments assessing the effect of O. lineata on C. vaginalis. Larvae were collected every day during the short term experiments. Metrics: Sponge Growth (SG); Larval Counts (LC); Brood Counts (BC); Brood chamber cross-sectional area (BA)...... 18
2. Results of two-factor ANOVA comparing change in disk diameter of O. lineata living in two species of sponge, C. vaginalis and N. digitalis. Sponges were preselected to either be brooding larvae or not brooding larvae...... 18
3. Results of ANOVA comparing growth, number of brood chambers, and number of larvae collected from C. vaginalis living with and without associated O. lineata. n is the number of sponge tubes with and without O. lineata for each experiment...... 19
4. Results from linear regression analyses and ANCOVA for number of larvae collected from C. vaginalis with and without associated O. lineata...... 19
5. Results of paired chemical cue choice assays in which O. lineata was presented sponge-conditioned seawater and filtered seawater, or seawater conditioned with C. vaginalis and N. digitalis. The percentage of choices for each treatment is given relative to the number of choices made (n) out of the total number of assays run. Asterisk denotes a significant difference from an expected equal probability of selecting either treatment using chi-squared goodness of fit...... 49
6. Choice experiments using chemical cues isolated from C. vaginalis and presented to brittlestars using different techniques and at various concentrations...... 50
7. Results from linear regression analyses and ANCOVA for number of gravid (a) and total (b) O. lineata found living in brooding (n=28) and non-brooding (n=18) C. vaginalis...... 51
8. Linear regressions based on morphological characteristics of O. lineata...... 77
9. Results from linear regression analyses and ANCOVA comparing wet mass and longest arm length between male and female O. lineata...... 77
10. Growth model parameters, residual sums of squares (RSS) and coefficient of determination (R2)...... 77
11. Comparison of community composition between sponges with and without associated large O. lineata using analysis of similarity (ANOSIM)...... 95
ix
12. Wet mass of Synalpheus hemphilli compared between C. vaginalis with and without associated large O. lineata and collection time (1 month, 2 month at NDR and 2 months at Dixie Shoals) using a two-factor ANOVA...... 95
x
LIST OF FIGURES
Figure Page
1. Percent change in disk diameter of O. lineata living in brooding and non- brooding C. vaginalis (n=4,3 respectively) and N. digitalis (n=3,3 respectively)...... 20
2. Mean percentage change in mass of C. vaginalis (±SE) living with and without associated O. lineata. There was no significant difference in any of the three trials (Table 3)...... 21
3. Mean number of brood chambers per 10 cm2 of C. vaginalis (±SE) after living with and without associated O. lineata. There was no significant difference in any of the three trials (Table 3)...... 21
4. Log-log plot of total larvae collected per day as a function of the number of brood chambers per cm2 of C. vaginalis. Fewer larvae were collected from sponge tubes with O. lineata (black diamonds) compared to sponges without associated O. lineata (open diamonds; Table 4)...... 23
5. Mean number of larvae collected per day (±SE) (a) and the coefficient of variation (b) collected during 6 experiments from tubes of C. vaginalis with and without associated O. lineata. There was no significant difference in the number of larvae collected per day between the two treatments in any of the 6 experiments (Table 2)...... 24
6. After 4 months, percentage change in disk diameter of O. lineata (bars ±SE) was significantly greater for O. lineata living in long (n=11) and short (n=12) C. vaginalis sponge tubes (ANOVA F1,21= 13.0683 p = 0.0016). Long tubes had significantly more brood chambers per tube (squares ±SE) than short tubes (ANOVA F1,21=7.35 p=0.013)...... 52
7. Mean percentage change in disk diameter of O. lineata (±SE) caged inside three sponge species for 4 months: Callyspongia vaginalis (n=2), Niphates digitalis (n=7), and Aplysina archeri (n=3). Despite low replication, mean growth of O. lineata in C. vaginalis was similar to that from previous experiments (see Fig. 6)...... 53
8. Size frequency distribution (bars) and the percent of mature brittlestars at each size class (diamonds) of the 204 O. lineata living inside 35 C. vaginalis...... 54
9. Frequency of observed ratios of male to female O. lineata living in a single C. vaginalis (n=35)...... 55
10. Abundance of gravid Ophiothrix lineata (a) and total abundance of O. lineata (b) as a function of surface area of C. vaginalis. Sponges with brood chambers (closed circle, solid line; n=28) and without brood chambers (open circle, dashed line; n=18). Regression and ANCOVA analysis in Table 7...... 56 xi
11. Development of O. lineata from embryo 16.5 hours after fertilization (A and B); 2 days and 21.5 hours (C); juvenile breaking free from fertilization membrane (fm) 6 days after fertilzation (D); and juvenile with pigment spots (ps) on the central disk (E)...... 78
12. log10 – log10 plot of longest arm length (squares; n=239) and calculated dry mass (circles; n=286) relative to disk diameter of O. lineata. Both characteristics are positively correlated to disk diameter and calculated dry mass is based on a regression of dry to wet mass of O. lineata (Table 8)...... 79
13. Size frequency distribution of male (n=93) and female (n=72) O. lineata...... 80
14. Walford plot of initial and final disk diameter from 44 O. lineata living in single tubes of C. vaginalis for 60-63 days. Predicted final disk size for each growth model is plotted along with a zero growth line...... 81
15. Disk diameter at age for four growth models from parameter estimates in Table 10...... 82
16. Non-metric multidimensional scaling ordinations of community structure in sponges with and without associated large O. lineata from North Dry Rocks after 1 month (a), and 2 months (b), Dixie Shoals after 1 month (c), and surveys of established C. vaginalis-associated community (d). Filled shapes represent sponges with large O. lineata and open shapes represent sponges without associated large O. lineata. Relative position in space denotes similarity/ dissimilarity, with points closer together being more similar...... 96
17. Average total abundance (±standard error) of associated fauna surveyed in established C. vaginalis. Sponges were grouped based on the presence of at least one large O. lineata > 6 mm disk diameter...... 97
18. Non-metric multidimensional scaling ordinations of community structure from established sponges (square) and immigrant communities (circle) after two months in the field. Dotted lines denote significant differences between established and immigrant communities based on ANOSIM (Table 1). Sites are noted above each point: North Dry Rocks (NDR), Dixie Shoals (DS) and Pickles Reef (PR)...... 98
19. Wet mass of Synalpheus hemphilli in surveyed C. vaginalis at two sites, North Dry Rocks (NDR) and Pickles Reef. There was no difference in wet mass between shrimp from C. vaginalis with or without associated large O. lineata...... 99
20. Non-metric multidimensional scaling ordinations of community structure for sponges after 1 month and 2 months at North Dry Rocks. Filled shapes represent sponges with large O. lineata and open shapes represent sponges without associated large O. lineata...... 100
xii
21. Average abundance of immigrant fauna (+standard error) found in C. vaginalis with and without associated large O. lineata. Tagged O. lineata were not included in counts...... 101
22. Average wet mass (±standard error) of Synalpheus hemphilli that immigrated to C. vaginalis with and without associated large O. lineata. Numbers denote the number of sponges with at least one Synalpheus hemphilli present. Both the presence of O. lineata and collection site/time had a significant effect on wet mass of immigrant Synalpheus hemphilli (Table 2)...... 102
xiii
CHAPTER 1: THE ASSOCIATION OF OPHIOTHRIX LINEATA AND CALLYSPONGIA VAGINALIS: IS THE BRITTLESTAR A PARASITE ON SPONGE LARVAE?
ABSTRACT
The Caribbean brittlestar Ophiothrix lineata has a species specific association with the tube
sponge Callyspongia vaginalis. Deposit-feeding by O. lineata on the surface of C. vaginalis
may benefit the sponge by increasing filtration efficiency. However, C. vaginalis also broods
larvae that are released year round and could provide another food resource to O. lineata. In lab
feeding assays, O. lineata readily consumed larvae from C. vaginalis and the vase sponge
Niphates digitalis. In the field, O. lineata grew faster when living in both species of brooding sponges than when living in non-brooding sponges. After 8 months, C. vaginalis with or without
brittlestars had no significant differences in the number of brood chambers in sponge tissue or in
overall sponge growth. Variation in larval output was high among tubes of C. vaginalis and
sponge tubes without O. lineata released more larvae and were more variable in larval output
relative to sponge tubes with O. lineata. In addition, there were significantly fewer larvae
relative to the number of brood chambers in C. vaginalis with O. lineata present. Previous
surveys have found O. lineata evenly distributed across both brooding and non-brooding C.
vaginalis. This even distribution of O. lineata across all C. vaginalis, as well as spatial and
temporal variation in larval release by C. vaginalis likely reduces encounter rates of O. lineata
with sponge larvae. Depending on the reproductive status of the sponge, the association between
O. lineata and C. vaginalis varies from a simple commensalism to a larval parasitism.
2
INTRODUCTION
Facilitative interactions between organisms, in which one member of an association benefits and the other is helped or unaffected, can be important in determining community
structure (Stachowicz 2001, Bruno et al. 2003). However, determining whether interactions are
facilitative is often difficult, because interspecific associations can fluctuate between
commensalism, mutualism, and even parasitism (Hoeksema & Bruna 2000, Thomson 2003, Hay
et al. 2004). Cleaning symbiosis, an often-cited example of mutualism on coral reefs, may
become detrimental to client fishes when ectoparasite abundance on clients is low and cleaning
fishes instead remove scales and mucus (Cheney & Cote 2005). Additionally, client fishes may
consume cleaner fishes, and the absence of cleaning stations may have little effect on the
ectoparasite abundance of clients (Cote 2000, Freckleton & Cote 2003). While interspecific
associations can be defined by the sum of the costs and benefits to participants (Bronstein 1994,
Hay et al. 2004), this requires a full understanding of the impacts to each participant with and
without the association.
Sponges are often a dominant component of the benthos, and are well known as hosts to a
taxonomically diverse population of organisms living on or inside them (Pearse 1949, Tyler &
Bohlke 1972, Ruetzler 1975, Pawlik 1983, Hendler 1984b, Duffy 1992). The tube sponge
Callyspongia vaginalis is one of the most common sponges on Caribbean reefs (Pawlik et al.
1995) and provides habitat for shrimps, amphipods, and brittlestars (Henkel & Pawlik 2005,
Rhyne & Lin 2006, Thomas & Klebba 2006). In Belize and the Florida Keys, one of the most
common inhabitants of C. vaginalis is the brittlestar Ophiothrix lineata which lives within tubes
of C. vaginalis (Kissling & Taylor 1977, Hendler 1984b, Henkel & Pawlik 2005). From within
the sponge osculum, the brittlestar deposit-feeds at night by extending its arms out over the outer
3
surface of the sponge tube (Hendler 1984a). The relationship between C. vaginalis and O.
lineata has been proposed by Hendler (1984) to be a mutualism: the brittlestar gains refuge and
a food source, while the sponge derives enhanced filtration efficiency from the cleaning activity
of the brittlestar. Hendler (1984) noted, however, that direct evidence of an advantage to the
sponge in filtration efficiency or increased growth rate remained to be demonstrated.
In addition to a food resource, O. lineata gains a physical refuge from predation within
the sponge (Hendler 1984a). The brittlestar has 99% fidelity in associating with C. vaginalis
(Henkel & Pawlik 2005) and uses chemical cues to detect the preferred sponge over other sponge
species (Chapt. 2). Growth of O. lineata is also greater for brittlestars living in C. vaginalis
compared to other sponge species (Chapt. 2). The high specificity of O. lineata for C. vaginalis
suggests an obligate relationship. Survival of the sponge, however, is not dependent on the
brittlestar, as 15% of C. vaginalis in Key Largo, FL did not contain any O. lineata (Henkel &
Pawlik 2005); and O. lineata is absent from C. vaginalis in the Bahamas Islands (Henkel pers.
obs.).
While the sponge-brittlestar association is facultative for the sponge, to classify the
association as mutualistic, enhanced fitness would be necessary, such as increased reproductive output or growth, in sponges having O. lineata. Callyspongia vaginalis is dioecious and broods larvae in distinct chambers randomly distributed in the sponge tubes. Free-swimming, parenchymellae larvae, 0.5-1.4 mm in length, are released during daylight hours throughout most of the year (Lindquist & Hay 1996, Lindquist et al. 1997). Larval trapping studies have reported 3 to over 200 larvae released by C. vaginalis during the day (Lindquist et al. 1997). The size of C. vaginalis larvae is at the upper range of the gut constituents of O. lineata examined by
4
Hendler (1984a), and considering the abundance of larvae, O. lineata may actually consume
larvae from C. vaginalis.
The potential for O. lineata to consume larvae from C. vaginalis poses the question: is
the association between O. lineata and C. vaginalis a mutualism, commensalism or parasitism?
To answer this question, I examined growth and reproduction in C. vaginalis living with and without O. lineata for up to 8 months on coral reefs off Key Largo, FL. Reproductive output
was quantified by collecting sponge larvae and assessing the number of brood chambers from
sponges living with and without O. lineata. I also conducted lab feeding assays by providing brittlestars with sponge larvae as well as field based experiments to assess differences in larval output from sponges with and without O. lineata. I further examined the growth of O. lineata living in brooding and non-brooding C. vaginalis and the brooding sponge Niphates digitalis to determine if the added larval food resource results in increased growth of O. lineata.
MATERIALS AND METHODS
Predation of sponge larvae by Ophiothrix lineata
Laboratory feeding assays were conducted to assess predation of sponge larvae by O. lineata. Larvae were collected from C. vaginalis and another brooding vase sponge, N. digitalis, using the technique described below for larval trapping experiments. Free swimming larvae from each sponge species were placed into shallow dishes with ~250 ml seawater and a single O. lineata. Dishes were placed in the dark for 5 hours, and the number of larvae remaining was compared to control dishes that did not have a brittlestar present. In addition, video of O. lineata consuming larvae from C. vaginalis was taken using a Sony HandyCam and the LED lighting from the camera.
5
An additional experiment examining sponge larval predation by O. lineata was conducted
at North Dry Rocks, FL (25o 07.850' N; 80o 17.521' W), a shallow 10 m subtidal patch reef, by comparing the growth of O. lineata living in brooding and non-brooding C. vaginalis and N.
digitalis. Sponge tubes, ~12 cm in height, were collected and 4-6 longitudinal slices were made
with a scalpel 5 cm from the top of the sponge to 5 cm from the base, noting the presence or
absence of brood chambers. To make single vases of N digitalis habitable for brittlestars
(Henkel and Pawlik 2005) the osculae of N. digitalis were constricted using monofilament line,
so that the oscular diameter was similar to the average diameter of C. vaginalis (~2.5 cm).
Sponges were allowed to heal for 2 weeks in the field. After the healing period, O. lineata were
collected from C. vaginalis on the reef and brought back to the lab. Initial disk diameter was
measured using digital calipers and brittlestars were tagged with a spot of the histological dye
Congo Red on their oral surface. Brittlestars were held overnight in a recirculating aquarium and
a single O. lineata was transplanted to each experimental sponge tube. After two months,
brittlestars and sponges were collected. Final disk diameter of tagged O. lineata was measured
and sponges were carefully dissected to determine the presence or absence of brood chambers.
Only O. lineata living in sponges and verified to be in their originally assigned sponge tubes
were used in analyses.
Effect of Ophiothrix lineata on Callyspongia vaginalis
Three long term experiments were conducted to examine the effects of the brittlestar O.
lineata on the host sponge C. vaginalis. The first experiment was conducted at North Dry Rocks
and then two additional experiments on top of the Aquarius undersea habitat at Conch Reef (24o
56.965'N 80o 27.224'W). Tubes of C. vaginalis, 12 cm in height, were collected from 10 m
6
depth on a patch reef at North Dry Rocks and brought back to the lab in containers of seawater.
Sponge tubes were weighed in seawater on an electronic balance and then immediately returned
to the field. Individual tubes were reattached upright to a brick or acrylic plate that was
previously attached to the substratum using a cable tie through the base of each tube. A single O.
lineata was haphazardly placed in half of the sponge tubes to create two treatments, with and
without associated O. lineata.
During the first experiment, sponges were inspected every 2 weeks to ensure the presence
or absence of O. lineata. Immigrant brittlestars were occasionally found in the without O. lineata treatment and these were removed from sponge tubes by prodding them with a long
hooked stick. The two subsequent experiments were conducted on top of the Aquarius habitat to
reduce effort spent monitoring for immigrant brittlestars. The Aquarius habitat, with its base in
~20 m seawater and the top of the platform at 9 m depth, provided a platform that would prevent
O. lineata from invading the experimental treatments. During the second Aquarius experiment,
sponges were enclosed in 4 separate 1.5 x 0.75 m cages made from 1.7 cm plastic mesh to reduce
the possible effects of predation on sponge growth. Both treatments were equally distributed in
each cage. In addition, 4 tubes of C. vaginalis with O. lineata present were placed outside of the
cage to assess the possible predation effect on sponge growth.
Larval traps were constructed using fine nylon mesh (pantyhose) placed over a ~12 x 8
cm cylindrical 1.7 cm plastic mesh frame. A 50 ml plastic centrifuge tube with tip cut off was
attached at one end of the nylon to create a funnel. A 200 ml plastic collection container, with
two windows covered by 50 µm mesh to permit some water flow through the container, was
placed over the plastic funnel. Larval collection bottles were centered over the osculum of the
sponge, with the plastic frame preventing direct contact of the larval traps with the sponge
7
surface and thereby minimizing any effects on the pumping of C. vaginalis or deposit-feeding by
O. lineata. For each long-term experiment, larval traps were placed over sponges for 2-5 days
and bottles were collected daily in the afternoon (Table 1). Larvae were quantified by counting
using a dissecting microscope.
After at least 6 months, sponge tubes were collected, brought back to the lab in seawater
and final wet weight measured. Oscular diameter and tube height were also measured; and inner
tube surface area was calculated using the equation for a cylinder. The presence or absence of
brood chambers was determined by carefully slicing sponge tubes longitudinally in ~7 mm
strips. This distance was enough to ensure brood chambers were only counted once. For
sponges from the 2006 and 2007 experiment, photographs were taken of brood chambers and
cross sectional area was measured using the image analysis program ImageJ v1.41h.
Three short term experiments were conducted to examine larval predation in the field
(Table 1). For the experiments conducted at Dixie Shoals (25o 04.66' N; 80o 18.74' W) and
Pickles Reef (24o 59.286' N; 80o24.600' W), sponge tubes, ~12 cm in height, were collected and
attached as described previously. Sponge tubes were allowed to heal for 7 days, and then a
single O. lineata was haphazardly placed into half of the sponge tubes and larval traps were
placed over all tubes.
The third experiment was conducted on the shipwreck USS Spiegel Grove (25° 4.000'N
80o 18.651'W). A single tube (~12 cm tall) of multi-tubed C. vaginalis was selected and a piece of fiberglass window screen was placed at the base of the selected tube, by slicing the tube base with a scalpel, placing the screen across the cut surface, and closing the cut surface with a cable tie. A single O. lineata was placed in half of the tubes, with the screen preventing the brittlestar from moving to other tubes within the multi-tubed sponge. Larval traps were placed over the
8
mesh-bottomed sponge tubes and bottles were collected daily in the afternoon with larvae
quantified as previously described.
Statistical analyses
Growth of O. lineata was calculated as percent change in disk diameter and compared
between brittlestars living in brooding and non-brooding sponges using a two-factor ANOVA.
The presence of brood chambers and sponge species, C. vaginalis and N. digitalis, were treated as fixed factors. The effect of O. lineata on growth of C. vaginalis was examined by comparing percentage change in mass for sponges with and without O. lineata using ANOVA for each long term experiment. Linear regression analysis was used to assess the relationship between the number of brood chambers in C. vaginalis and the total cross sectional area of brood chambers in a sponge tube. The number of brood chambers per 10 cm2 sponge tissue was also compared
between treatments using ANOVA on log-transformed data. Total number of larvae collected
per day from sponges with and without associated O. lineata was compared for each long and
short term larval trapping experiment using ANOVA on log-transformed data. Predation of
sponge larvae by O. lineata would reduce not only the total number of larvae collected, but also
reduce the variation in the number of larvae collected between sponges with O. lineata present
compared to sponges without O. lineata. This hypothesis was examined by comparing the
coefficient of variation of the average larvae captured per day from the 6 field experiments using
a one-tailed paired t-test.
Larval output was also compared between C. vaginalis with and without O. lineata using
ANCOVA on combined data from the North Dry Rocks, 2 Aquarius, and Dixie Shoals
experiments. This subset of data was used as information on the number of brood chambers in
9
sponges was only available for these experiments. The number of larvae collected per day was
the response variable and the number of brood chambers per 10 cm2 sponge tissue was the
covariate. Brood chambers were quantified at the end of experiments; therefore, only larval
counts from the same time-period were used in analyses, as the abundance of brood chambers may have changed over time and may not be related to previous larval collections. In addition, only brooding sponges or sponges that had produced larvae were included in larval and brood chamber analyses. All statistics, including tests for the assumptions of ANOVA and ANCOVA
(Sokal & Rohlf 1981) were calculated using JMP 7.0 (SAS Institute).
RESULTS
An average of 64% of larvae placed in dishes with O. lineata were consumed within 5 hours relative to dishes with no brittlestar present. Growth of O. lineata did not differ between sponge species (Table 2) and was greater in both brooding C. vaginalis and N. digitalis (18.2%
±6.0 and 29.0% ±7.7 SE, respectively; Fig. 1) relative to brittlestars living in non-brooding C.
vaginalis and N. digitalis (9.5% ±6.9 and 10.5% ± 4.7, respectively; p=0.065 Table 2).
There was no statistical difference in percentage change in tissue mass of C. vaginalis
after at least 6 months living with and without associated O. lineata (Table 3; Fig. 2). Sponge
tubes had an overall loss of tissue during the uncaged 2006 Aquarius experiment (Fig 2). In the
2007 Aquarius experiment, the 4 uncaged tubes of C. vaginalis with associated O. lineata grew
to a similar extent as caged sponges (8.9% ±4.8SE and 9.5% ±3.4SE, respectively).
In experimental sponges that had brood chambers present, the number of brood chambers per tube was positively correlated to total brood chamber cross-sectional area of the sponge tube
(log10 number of brood chambers = 1.35 log10 total brood cross-sectional surface area + 0.165;
10
R2=0.9223 p<0.001). There was no difference in the number of brood chambers per 10 cm2
sponge tissue between sponges living with and without O. lineata (Table 3; Fig. 3).
In experiments using larval traps over sponges, the number of larvae captured per day
increased with increasing abundance of brood chambers, and the number of larvae captured per
day was significantly greater in sponges without O. lineata compared to sponges with O. lineata
(Table 4, Fig 4). When comparing larvae released during the 6 larval trapping experiments, the
number of larvae captured per day from sponges without O. lineata was consistently greater than
from sponges with O. lineata, although not significantly greater (Table 3; Fig. 5a). The
coefficient of variation of larvae captured per day was high (>1) for all treatments and
significantly less in sponges with O. lineata present compared to sponges without O. lineata
present (Fig. 5b; one-tailed paired t-test: t15=2.018 p=0.0498).
DISCUSSION
The results of this study suggest that the association between the brittlestar Ophiothrix
lineata and the tube sponge Callyspongia vaginalis is not necessarily a mutualism, but may vary
from a simple commensalism with the brittlestar living in and feeding on the surface of the sponge having no effect on the sponge, to a parasitism with the brittlestar consuming larvae produced by the sponge. The presence of O. lineata did not result in increased growth or reproductive effort in C. vaginalis, and the brittlestar would readily eat larvae produced by the sponge. While other sponge-dwelling fauna directly consume sponge tissue (Pawlik 1983, Rios
& Duffy 1999), this study is the first to describe predation of sponge larvae by a sponge symbiont. Previous studies have found larvae of C. vaginalis to be chemically defended against
11
planktivorous predators (Lindquist & Hay 1996). The present study, however, suggests that
larval predators may be dwelling within the sponge itself.
In three separate field experiments, there was no effect of O. lineata on the growth or
reproductive potential of C. vaginalis. After 8 months of living with or without the associated
brittlestar, C. vaginalis tubes exhibited no consistent pattern with respect to either growth (Fig.
2) or abundance of brood chambers (Fig. 3). Sponge growth can be affected by water flow
(Kaandorp 1999), and the high variation observed in growth of C. vaginalis may be a function of
transplantation and small scale differences in flow for each sponge tube. The loss of sponge
tissue during the 2006 Aquarius experiment could be due to a variety of factors including food
availability or predation pressure. Four tubes of C. vaginalis left uncaged on the Aquarius in
2007 had similar growth to caged sponges suggesting predation pressure was not responsible for the loss of tissue in 2006. However the two experiments were conducted during different
seasons (2006-Winter-Spring, 2007 Spring-Summer) and may represent seasonal variation in
predation pressure or sponge growth. Interestingly, larval output and brood chamber abundance
was similar in 2006 as in other experiments, despite the loss of mass (Fig. 4 and 5a).
Based on laboratory feeding assays, O. lineata is clearly able to consume free swimming sponge larvae in a confined environment. Videography with infrared illumination revealed that brittlestars trap larvae with their tube feet, pass them along an arm to the mouth and then consume them, often several larvae at a time. Larvae were observed escaping tube feet and occasionally swimming out of the mouth of brittlestars; however, sponge larvae did not appear to avoid O. lineata. Larval predation in the laboratory corroborates the results of field experiments in which O. lineata grew more in brooding vs. non-brooding C. vaginalis and N. digitalis (Fig.
1). Replication in these field experiments was low because of loss of replicates from storm
12
events. Although variation was great in all treatments, O. lineata living in brooding C. vaginalis
and N. digitalis grew more, suggesting that the additional larval food resource provided by
brooding sponges enhanced brittlestar growth.
The effect of predation on sponge larvae by O. lineata is also evident by differences in
the number of larvae collected in larval traps from sponges with and without associated
brittlestars Although the large variation in larval output among sponge tubes resulted in no
statistical difference between treatments, on average more larvae were collected from C. vaginalis without O. lineata in 5 of the 6 larval trapping experiments, with means ranging from
24-241% more larvae captured per day relative to sponges with O. lineata (Fig. 5a). The pattern of lower numbers of larvae collected from sponges with associated brittlestars is based on data collected across three seasons with samples from 99 sponges with O. lineata and 105 without the associated brittlestar. Using a subset of these experiments for which data on the abundance of brood chambers was also available, the number of larvae captured per day was significantly greater in sponges without brittlestars (Fig. 4). Additionally, sponges without brittlestars had a higher coefficient of variation compared to sponges with brittlestars (Fig. 5b), suggesting that larval predation by O. lineata reduced the variability in the number of larvae released and may constrain the maximum larval output by the sponge. The patterns of reduced larval output and lower variation in larval production by sponges with associated brittlestars, along with increased growth of O. lineata living in brooding sponges (Fig. 1), support the hypothesis that O. lineata consumes C. vaginalis larvae in the field. Considering that O. lineata exhibits strong host specificity for C. vaginalis (Henkel and Pawlik 2005, Chapt. 2), has no positive effect on the growth of the sponge, and may reduce sponge fitness by consuming sponge larvae, the association between O. lineata and brooding C. vaginalis can be considered a parasitism.
13
Pulses of larval release by C. vaginails may provide a mechanism whereby sponge larvae escape predation by O. lineata. Among sponges on which larval traps were deployed, there was no clear pattern to the number of larvae captured on any single day or from any single sponge, although sponges that released a large number of larvae (>100) typically released fewer the following days. Larval release was most likely based on development of individual brood chambers within a sponge and not synchronized for the entire sponge. The maximum number of larvae captured in a single day was 437 from a C. vaginalis without an associated brittlestar and
147 from a sponge with a brittlestar living inside. The release of large numbers of sponge larvae may inundate the brittlestar and reduce capture efficiency. In addition, the more frequently observed low levels of larval output (1-10 larvae per day from sponges with and without O. lineata) may be more difficult for O. lineata to detect and reduce larval predation pressure.
Timing of larval release may also benefit the sponge, because C. vaginalis releases larvae during the late morning and afternoon (Lindquist et al. 1997). Deposit-feeding by O. lineata occurs almost entirely at night, although brittlestars have been observed deposit-feeding during the day (Hendler 1984b). Larger O. lineata are typically further down in sponge tubes during the day, with arms up against tube walls. Release of larvae by C. vaginalis has been attributed to the use of photic cues in larval settlement (Lindquist et al. 1997). Daylight spawning may have the added benefit of occurring during reduced activity of O. lineata. The multi-tubed morphology of
C. vaginalis may also reduce larval predation, given that brood chambers are scattered throughout the sponge wall and there is no discernable timing to larval release from individual brood chambers, reducing the likelihood of O. lineata being present when larvae are released. In addition to spatial variation within a sponge, surveys of 46 C. vaginalis with associated O. lineata found only 61% of sponges had brood chambers, suggesting that O. lineata does not
14
preferentially select female C. vaginalis (Chapt. 2). The relatively even distribution of O. lineata
across male and female C. vaginalis may further reduce the effect of larval predation on the
sponge.
The discovery that O. lineata is a larval parasite of C. vaginalis raises interesting
questions about the fitness impact of the brittlestar on its host sponge. Based on data from all 6
larval trapping experiments, there was an average of 5.4 larvae released per day from sponge
tubes with associated brittlestars (n=99) compared to 7.4 larvae per day from sponges without
brittlestars (n=105). Thus, sponge tubes without brittlestars may release an average of 27% more
larvae relative to sponges with an associated brittlestar. The overall effect on populations of C. vaginalis would be determined by the relative population density of O. lineata. Densities of O.
lineata increase with the size of individual C. vaginalis (Chapt. 2), increasing the probability a
single O. lineata encountering sponge larvae. A previous survey in the Florida Keys found 85%
of C. vaginalis had O. lineata present, with only 58% of sponges containing brittlestars with a
disk diameter ≥ 5 mm (Henkel & Pawlik 2005). Smaller brittlestars live on the outside of sponge
tubes, while the larger size class lives inside C. vaginalis (Hendler 1984b, Henkel & Pawlik
2005), where they could prey on larvae. If O. lineata is associated with 58% of C. vaginalis and
61% of inhabited sponges are brooding, then larval predation could occur in 35% of C. vaginalis.
While the factors limiting population size of O. lineata are unclear, the presence of large O. lineata in only 58% of C. vaginalis suggests that sponge habitat is not limiting (Henkel & Pawlik
2005). Based on the data presented here, I would expect any increase in populations of O. lineata could have a negative impact on C. vaginalis; and the factors currently limiting population size of O. lineata may have an indirect positive effect on C. vaginalis by limiting the larval parasite.
15
LITERATURE CITED
Bronstein JL (1994) Our current understanding of mutualism. Quarterly Review of Biology 69:31-51
Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation into ecological theory. Trends In Ecology & Evolution 18:119-125
Cheney KL, Cote IM (2005) Mutualism or parasitism? The variable outcome of cleaning symbioses. Biology Letters 1:162-165
Cote IM (2000) Evolution and ecology of cleaning symbioses in the sea. Oceanography and Marine Biology, Vol 38 38:311-355
Duffy JE (1992) Host use patterns and demography in a guild of tropical sponge-dwelling shrimps. Marine Ecology Progress Series 90:127-138
Freckleton RP, Cote IM (2003) Honesty and cheating in cleaning symbioses: evolutionarily stable strategies defined by variable pay-offs. Proceedings of the Royal Society of London 270:299-305
Hay ME, Parker JD, Burkepile DE, Caudill CC, Wilson AE, Hallinan ZP, Chequer AD (2004) Mutualisms and aquatic community structure: The enemy of my enemy is my friend. Annual Review of Ecology Evolution and Systematics 35:175-197
Hendler G (1984a) The association of Ophiothrix lineata and Callyspongia vaginalis a brittlestar- sponge cleaning symbiosis? Marine Ecology 5:9-28
Hendler G (1984b) The association of Ophiothrix lineata and Callyspongia vaginalis: a brittlestar-sponge cleaning symbiosis? Marine Ecology 5:9-28
Henkel TP, Pawlik JR (2005) Habitat use by sponge-dwelling brittlestars. Marine Biology 146:301-313
Hoeksema JD, Bruna EM (2000) Pursuing the big questions about interspecific mutualism: a review of theoretical approaches. Oecologia 125:321-330
Kaandorp JA (1999) Morphological analysis of growth forms of branching marine sessile organisms along environmental gradients. Marine Biology 134:295-306
Kissling DL, Taylor GT (1977) Habitat factors for reef-dwelling ophiuroids in the Florida Keys. Proceedings int Coral Reef Symp 3:225-231
Lindquist N, Bolser R, Laing K (1997) Timing of larval release by two Caribbean demosponges. Marine Ecology Progress Series 155:309-313
Lindquist N, Hay ME (1996) Palatability and chemical defense of marine invertebrate larvae. Ecological Monographs 66:431-450
16
Pawlik JR (1983) A sponge eating worm from Bermuda: Branchiosyllis oculata (Polychaeta Syllidae). Marine Ecology 4:65-80
Pawlik JR, Chanas B, Toonen RJ, Fenical W (1995) Defenses of Caribbean sponges against predatory reef fish. I. Chemical deterrency. Marine Ecology Progress Series 127:183-194
Pearse AS (1949) Notes on the inhabitants of certain sponges at Bimini. Ecology 31:149-151
Rhyne AL, Lin J (2006) A western Atlantic peppermint shrimp complex: Redescription of Lysmata wurdemanni, description of four new species, and remarks on Lysmata rathbunae (Crustacea : Decapoda : Hippolytidae). Bull Mar Sci 79:165-204
Rios R, Duffy JE (1999) Description of Synalpheus williamsi, a new species of sponge-dwelling shrimp (Crustacea: Decapoda: Alpheidae), with remarks on its first larval stage. Proceedings of the Biological Society of Washington 112:541-552
Ruetzler K (1975) Ecology of Tunisian commercial sponges. Tethys 7:249-264
Sokal RR, Rohlf FJ (1981) Biometry: the principles and practice of statistics in biological research, Vol. W. H. Freeman, San Francisco
Stachowicz JJ (2001) Mutualism, facilitation, and the structure of ecological communities. Bioscience 51:235-246
Thomas JD, Klebba KN (2006) Studies of commensal leucothoid amphipods: Two new sponge- inhabiting species from South Florida and the Western Caribbean. Journal of Crustacean Biology 26:13-22
Thomson J (2003) When is it mutualism? American Naturalist 162:S1-S9
Tyler JC, Bohlke JE (1972) Records of sponge dwelling fishes primarily of the Caribbean. Bull Mar Sci 22:601-642
17
Table 1: Timing of long and short term experiments assessing the effect of O. lineata on C. vaginalis. Larvae were collected every day during the short term experiments. Metrics: Sponge Growth (SG); Larval Counts (LC); Brood Counts (BC); Brood chamber cross-sectional area (BA).
Location Depth Dates Metrics Larval Traps Long Term North Dry Rocks 10m Feb 3-Oct 4, 2005 SG, LC, BC Jun 20-25, Jul 5-6, Aug 8-14 Aquarius, Conch Reef 10m May 31-Nov 12, 2006 SG, LC, BC, BA Jul 2-6, Nov 8-12 Aquarius, Conch Reef 10m Nov 17 2006-Jul 6, 2007 SG, LC, BC, BA Jul 2-6 Short Term Dixie Shoals 12m May 20-25 2005 LC, BC Spiegel Grove 25-30m Oct 3-5 2007 LC Pickles Reef 12m Dec 4-8 2007 LC
Table 2: Results of two-factor ANOVA comparing change in disk diameter of O. lineata living in two species of sponge, C. vaginalis and N. digitalis. Sponges were preselected to either be brooding larvae or not brooding larvae.
Source DF F Ratio P Brood Chambers 1 4.406 0.065 Species 1 0.840 0.383 Brood Chambers x Species 1 0.580 0.466
18
Table 3: Results of ANOVA comparing growth, number of brood chambers, and number of larvae collected from C. vaginalis living with and without associated O. lineata. n is the number of sponge tubes with and without O. lineata for each experiment.
Treatment Experiment n df F P percent change in mass of C. vaginalis North Dry Rocks 16/12 1,27 0.356 0.556 2006 Aquarius 22/24 1,45 0.168 0.068 2007 Aquarius 15/10 1,24 1.890 0.183 log(number of brood chambers per 10 cm2 tissue) North Dry Rocks 15/11 1,25 0.995 0.329 2006 Aquarius 17/22 1,38 0.011 0.919 2007 Aquarius 2/2 log (number of larvae captured per day) Dixie Shoals 18/18 1,35 0.421 0.521 North Dry Rocks 18/17 1,34 1.617 0.212 2006 Aquarius 21/27 1,47 1.427 0.238 2007 Aquarius 8/3 1,10 0.738 0.413 Spiegel Grove 16/18 1,33 0.447 0.508 Pickles Reef 18/22 1,39 0.521 0.475
Table 4: Results from linear regression analyses and ANCOVA for number of larvae collected from C. vaginalis with and without associated O. lineata.
Treatment Equation R2 Source df F P log total larvae per day+1 (Y) vs. log brood chambers per cm sponge tissue +1 (X) With O. lineata Y = 4.832 X + 0.261 0.222* Treatment 1 3.943 0.0498 Without O. lineata Y = 2.811 X + 0.227 0.127** Slope 1 1.491 0.2250 * p≤0.05 **p<0.001\
19
40%
35% Brooding Non-Brooding 30%
25%
20%
15%
10%
PERCENT CHANGE IN DISK DIAMETER CHANGE PERCENT 5%
0% C. vaginalis N. digitalis
Fig. 1: Percent change in disk diameter (±SE) of O. lineata living in brooding and non-brooding C. vaginalis (n=4,3 respectively) and N. digitalis (n=3,3 respectively).
20
35%
30% With
Without 25%
20%
15%
10%
PERCENT CHANGE MASS PERCENT CHANGE 5%
2006 Aquarius 0%
2005 Dry Rocks 2007 Aquarius -5%
-10%
-15%
Fig. 2: Mean percentage change in mass of C. vaginalis (±SE) living with and without associated O. lineata. There was no significant difference in any of the three trials (Table 3).
21
5
4.5 With TISSUE
Without 4 SPONGE
2 3.5 cm
10 3 PER
2.5
2 CHAMBERS
1.5 BROOD 1 OF
0.5
NUMBER 0 2005 Dry Rocks 2006 Aquarius 2007 Aquarius
Fig. 3: Mean number of brood chambers per 10 cm2 of C. vaginalis (±SE) after living with and without associated O. lineata. There was no significant difference in any of the three trials (Table 3).
22
1.8 With
1.6 Without
1.4
1.2
1
0.8
0.6 log (Total Larvae log (Total Larvae per Day+1)
0.4
0.2
0 0 0.05 0.1 0.15 0.2 log (Brood Chambers per Sponge Area +1)
Fig. 4: Log-log plot of total larvae collected per day as a function of the number of brood chambers per cm2 of C. vaginalis. Fewer larvae were collected from sponge tubes with O. lineata (black diamonds) compared to sponges without associated O. lineata (open diamonds; Table 4).
23
a. 45
40
35
30 With Without 25
20
15
10
Mean Larvae Collected per Day 5
0
3 b.
2.5
2
1.5
1
Coefficient of Variation 0.5
0 2007 Grove 2006 2007 Spiegel Rocks Reef Shoals Aquarius Aquarius 2005 Dry 2005 Dixie 2007 Pickles
Fig. 5: Mean number of larvae collected per day (±SE) (a) and the coefficient of variation (b) collected during 6 experiments from tubes of C. vaginalis with and without associated O. lineata. There was no significant difference in the number of larvae collected per day between the two treatments in any of the 6 experiments (Table 3); however, the coefficient of variation was significantly less in sponges with O. lineata compared to sponges without O. lineata (one-tailed paired t-test: t15=2.018 p=0.0498).
24
CHAPTER 2: HOST SPECIALIZATION AND LIMITATION OF AN OBLIGATE SPONGE-DWELLING BRITTLESTAR
25
ABSTRACT
Host specificity is a common phenomenon, but the benefits of specifity are often difficult
to discern. Defining adaptations that allow exploitation of a single host can provide clues to the evolution of specific associations. On coral reefs off the Florida Keys, the sponge-dwelling brittlestar Ophiothrix lineata lives almost exclusively in the tube sponge Callyspongia vaginalis.
I examined chemotactic recognition by O. lineata to assess sponge host preferences using a y-
tube assay chamber. Relative to seawater controls, O. lineata preferentially selected seawater
conditioned by the preferred host sponge C. vaginalis and showed no preference for seawater conditioned by the infrequent host Niphates digitalis or non-host Aplysina archeri. When offered seawater conditioned by C. vaginalis and N. digitalis, O. lineata chose C. vaginalis 78% of the time. Field manipulations were performed to examine the impact of habitat size on growth of O. lineata living in C. vaginalis and to measure growth of O. lineata confined to live in all three sponge species. Growth of O. lineata was significantly greater in longer tubes of C. vaginalis than in shorter tubes (12 and 6 cm tall, respectively), and while O. lineata was able to survive in A. archeri, growth was greater for brittlestars living in C. vaginalis and N. digitalis.
Surveys of O. lineata in C. vaginalis revealed that 74.3% of C. vaginalis had at least one male and female O. lineata, with 37.1% of sponges having a greater proportion of males, both of which may increase brittlestar fertilization success. Abundance of O. lineata increased with
sponge size and did not differ between brooding and non-brooding C. vaginalis. The evolution
of the obligate association of O. lineata with C. vaginalis likely resulted as a consequence of
some combination of host abundance, enhanced food availability and greater probability of
mating success in the multi-tubed sponge.
26
INTRODUCTION
The development and maintenance of host specificity has long been recognized by evolutionary ecologists as an important means of speciation in terrestrial plant-herbivore and
host-parasite interactions (Giorgi et al. 2004; Hufbauer and Via 1999; Tompkins and Clayton
1999). Multiple factors can increase host specificity, including limited dispersal or interaction
with potential hosts, adaptive specialization resulting from some fitness gain from an association,
adaptive limitation resulting in lower fitness on alternative hosts, or increased probability of
finding a mate in a specific area (Agosta 2008; Combes 1991; Sotka 2005; Timms and Read
1999). The importance of host specificity in marine systems is less understood (Sotka 2005) and
ecologists have recently increased their focus on the foregoing mechanisms to describe the
impact of host specialization on speciation in marine systems (Faucci et al. 2007; Johnston and
Miller 2007; Sotka 2007).
Studies of host specificity in marine systems frequently cite plant-insect analogs, described as associations between small invertebrates with limited mobility or dispersal that associate with larger hosts for refuge and food (Hay et al. 1987; Sotka 2005). One of the best
described examples of host-mediated speciation in a marine system is that of alpheid shrimp that
live in association with sponges (Duffy 1996a; Duffy 1996b; Macdonald et al. 2006). In a
collection of 27 Synalpheus spp. from inside sponges on coral reefs in Belize, 55% were found in
only one species of sponge (Macdonald et al. 2006). The high level of host specialization in
Synalpheus is most likely a function of space partitioning based on sponge morphology, as well
as competitive interactions between shrimp species, the eusocial behavior of several species, and
the direct development and short dispersal of larvae (Duffy 1992; Duffy 1996b; Macdonald et al.
2006).
27
The pattern of host use observed in alpheid shrimp is probably one of many, considering
the abundance and diversity of sponges and their associated fauna (reviewed in Wulff 2006).
The sponge-dwelling brittlestar Ophiothrix lineata is a species-specific obligate living in
association with the tube sponge Callyspongia vaginalis (Henkel and Pawlik 2005). Although
O. lineata has been observed living in association with other sponges at low frequency (Hendler et al. 1995; Henkel and Pawlik 2005; Kissling and Taylor 1977), surveys of sponges in the
Florida Keys recorded O. lineata living in C. vaginalis 99% of the time, despite the presence of similarly shaped sponges that could provide refuge (Henkel and Pawlik 2005). The association between O. lineata and C. vaginalis has been described as a cleaning mutualism, with O. lineata deposit-feeding on the surface of the sponge (Hendler 1984). Comparisons of growth and larval output of C. vaginalis with and without associated O. lineata demonstrated that the sponge does not benefit from O. lineata (Chapt. 1). Furthermore, O. lineata is capable of consuming larvae released by C. vaginalis in lab-feeding experiments and may be a larval parasite on the sponge
(Chapt. 1). The association between O. lineata and C. vaginalis resembles the better studied plant-insect interactions, and provides a marine system for examining mechanisms driving selection for an obligate association.
Enhanced fitness is often inferred as the selection mechanism for a high degree of host specificity, although limited interactions with alternate hosts can also result in the appearance of host preferences (Tompkins and Clayton 1999). The presence of a species-specific chemical cue used in host recognition is evidence of a specialized adaptation that may reflect a fitness advantage. Echinoderms rely on chemical cues to recognize prey items (Sloan and Campbell
1982) as well as associated hosts (Clavico et al. 2006; Fourgon et al. 2007). The facultative sponge-dwelling brittlestar Ophiactis savigyni uses chemical cues to detect the sponge Geodia
28
corticostylifera (Clavico et al. 2006), but this study only examined the chemical cues of a single
potential host. In the Florida Keys, O. lineata lives primarily in association with C. vaginalis, and also lives in low frequency (< 1%) within the vase sponge Niphates digitalis (Henkel and
Pawlik 2005) and the tube sponge Verongula (Aplysina) lacunosa (Kissling and Taylor 1977). If the observed preference for C. vaginalis results in enhanced fitness for O. lineata, then I hypothesize that the brittlestar should distinguish C. vaginalis from other sponge hosts.
Host specificity imposes potential population limitations if host availability or size is limited. Size-specific shelter requirements limit crustacean populations through mortality, emigration, or stunting of the effected size classes (Beck 1995; Caddy 1986; Caddy and
Stamatopoulos 1990). Alpheid shrimps chose sponge habitat based on the size of suitable channels in the sponge tissue (Duffy 1992). In the Florida Keys, 69% of large O. lineata, > 5 mm disk diameter, lived in sponge tubes > 70 cm2 (Henkel in prep). Large O. lineata (≥ 5 mm
disk diameter) live inside sponge tubes as a refuge from fish predators (Hendler 1984), and
brittlestar abundance may be limited by available sponge habitat (Henkel and Pawlik 2005;
Kissling and Taylor 1977). Food availability, including surface area for deposit-feeding and
predation on sponge larvae, is a function of sponge size; therefore, limited food resources from
smaller sponges may also restrict the distribution of large O. lineata to larger tubes.
In the present study, I explored both specific adaptations and fitness advantages to host
exploitation by the brittlestar O. lineata. A choice assay was used to determine whether O.
lineata was capable of chemotaxis in response to chemical cues from three species of sponge:
Callyspongia vaginalis, Niphates digitalis, and Aplysina archeri. The last of these was chosen
because it has a similar oscular diameter to C. vaginalis, but O. lineata has not been found living
in A. archeri (Henkel and Pawlik 2005). Because fitness advantages to host specificity are often
29
difficult to determine, controlled laboratory experiments are often employed to track growth and
reproduction of species living on alternate hosts over multiple generations (Sotka 2005).
Sponges are difficult to maintain in the lab; therefore, I used a field-based approach to examine
the fitness impacts of different sponge hosts by measuring the growth of O. lineata living in all
three sponge species. Focusing on the O. lineata–C. vaginalis association, I also examined the
potential fitness constraints of habitat size by comparing the growth of O. lineata living in two
different sized tubes of C. vaginalis. Finally, I compared and surveyed the sex distribution of O.
lineata living in C. vaginalis for insights into reproductive constraints on an obligate commensal.
MATERIALS AND METHODS
Host Choice Experiments
Three species of sponges, C. vaginalis, N. digitalis, and A. archeri, were collected using
SCUBA from shallow reefs off Key Largo, FL: North Dry Rocks, (25o 07.850 N; 80o 17.521W) and Pickles Reef (24 o 59.286 N; 80o24.6 W). Sponges were cut underwater, placed in mesh
cages attached to the substratum and allowed to heal for at least 2 days in the field. Cut sponges
were then transported in containers of seawater to the lab and kept in a large recirculating 100 l
holding tank. Water changes were conducted daily and sponges were used in experiments before
any deterioration in sponge health was observed (< 7 days). Individual O. lineata were collected
from C. vaginalis on the reef and held in perforated plastic containers submerged in a separate
aquarium using the same recirculating seawater as the sponges.
Seawater for chemical cue assays was collected over shallow coral reefs (~ 10 m deep)
from 2-3 m below the surface using a submersible pump, then filtered through a 500 µm filter-
fiber bag. Sponge-conditioned seawater was prepared by placing 1 L of living sponge tissue in 6
30
L of filtered seawater in a bucket with an aerator for 4 hours. Sponge volume was determined by volumetric displacement of seawater. Control seawater was prepared in the same manner, without the addition of sponge tissue. During the seawater incubation process, brittlestars were held in an aquarium containing filtered seawater that had not been exposed to sponges.
An assay chamber was constructed using tygon tubing with an inner diameter of 2.54 cm.
Two 14 cm long pieces of tube were connected to a 120o PVC Y-connector, with a shorter 6 cm long piece of tubing attached to the base of the Y that served as a brittlestar entrance. Two L- shaped fittings attached to 6 cm pieces of vertical tubing served as inlets for treatment seawater, and these were connected to the 20 cm pieces of tubing. Two MityFlex peristaltic pumps (Model
907-014) with 6.3 mm diameter peristalsis tubing delivered treatment and control seawater at a rate of 48 ml per minute. The assay chamber was placed in a shallow pan of control seawater and allowed to drain freely while water was added by the peristaltic pumps.
Three choice experiments were conducted using filtered seawater and seawater conditioned with one of the following sponge species: C. vaginalis, N. digitalis, and A. archeri.
In addition, choice assays were conducted using seawater conditioned with C. vaginalis and N. digitalis. Choice experiments were performed in the dark, as brittlestars are negatively photo- sensitive. A single O. lineata was placed into the assay pan and allowed to acclimate for 2 minutes. After the acclimation period, the brittlestar was placed inside the entrance tube and the pumps were activated. A choice occurred when the oral disk of the brittlestar passed beyond the
PVC Y-connector into one of the two treatment arms. Each brittlestar was given 10 minutes to make a choice between the two treatments, and was used only once in any paired experiment.
The assay chamber was flushed with control seawater between replicate assays, and the treatments were alternated between the two pumps and treatment arms to control for flow
31 differences. Preferences for a particular treatment were determined using a chi-square goodness of fit to test for deviation from an expected equal probability of selecting either treatment. Only replicates in which a brittlestar made a choice were used in analyses. A right-tailed Fisher’s exact test was used to determine if the probability of O. lineata making a choice is greater when
C. vaginalis is present than when it is absent. Choice experiments were further used to attempt to isolate the chemical cues from C. vaginalis responsible for host choice by O. lineata (Table 6).
Assays were conducted in Wilmington, NC using O. lineata collected from Key Largo, FL and kept in aquaria with artificial seawater (Red Sea brand) prepared to a salinity of 35 ppt. Crude organic extracts were prepared from C. vaginalis tissue collected from Key Largo, FL using standard extraction techniques (Pawlik et al. 1995). Paired choice experiments were conducted following the same choice protocol, using artificial seawater and treatment-conditioned seawater.
Treatments included organic extracts incorporated into gel matrix (as per Engel and Pawlik
2000) and resuspended extracts directly added to seawater. When choice experiments with extract treatments were unsuccessful, seawater conditioned with frozen whole sponge tissue was used in experiments (Table 6).
Effect of Habitat Size on Growth of Ophiothrix lineata
Growth of Ophiothrix lineata was monitored from March 11 to July 12, 2005 on a shallow patch reef at North Dry Rocks using SCUBA. To determine whether disk diameter of O. lineata correlates with overall body size, the mass of 143 towel dried O. lineata was determined and compared with disk diameter. While both metrics can be obtained non-lethally, disk diameter is not impacted by loss of arms due to handling damage or sub-lethal predation. Forty tubes of C. vaginalis, ~ 2 cm in oscular diameter, were cut for two habitat size treatments. Long
32
tubes were 12 cm and short tubes were 6 cm in height, representing ~ 35 and 75 cm2 inner tube
surface area, respectively. Long and short sponge tubes were cable-tied upright to bricks that
were secured to coral pavement and placed haphazardly on the reef, ≥ 2 m apart at 10 m depth
and allowed to heal for one week. Ophiothrix lineata, 5.5-8.0 mm disk diameter, were collected
from nearby sponges, brought back to the lab in seawater and their oral surfaces were tagged
with small dots of the histological dye Congo Red. Tagged brittlestars were kept in recirculating
seawater overnight to confirm survival and quality of tags. The following day, initial disk
diameter, measured as the distance from the base of one arm to the point on the disk directly
opposite, was measured to the nearest 0.5 mm using calipers. Tagged brittlestars were then
transplanted to the field and haphazardly assigned to one of the two sponge tube height
treatments (n=20 for each treatment). After 4 months, brittlestars and sponge tubes were
collected, brought back to the lab, and final disk diameter was measured. Sponge tubes were dissected longitudinally into 1 cm strips, and the number of brood chambers per tube were counted in order to determine whether habitat size is a proxy for potential sponge larval food resources. This width ensured that brood chambers were only counted once, as brood chambers did not exceed 1 cm width. Percentage change in the disk diameter of O. lineata was calculated and differences between the two habitat size treatments were compared using an ANOVA (Sokal and Rohlf 1981). Differences in the total number of brood chambers between long and short sponge tubes were compared using ANOVA on log10 transformed data. All statistics were
performed using JMP ver 7.0 (SAS Insitute).
Growth of Ophiothrix lineata Living on Multiple Sponge Hosts
33
Based on previous emigration studies, O. lineata will abandon sponges other than C. vaginalis within 24 hours (Henkel and Pawlik 2005); therefore, I devised a method to cage O. lineata inside sponge tubes to assess growth of the brittlestar in different sponge hosts. I first
examined potential caging effects by comparing the growth of O. lineata living in caged and
uncaged C. vaginalis at North Dry Rocks. A small piece of flexible plastic 5 mm mesh was
placed over the osculum of equally sized (10 cm tube height) C. vaginalis tubes, and tubes were
cable-tied upright to bricks that were secured to coral pavement and placed haphazardly on the
reef. Mesh prevented brittlestars from escaping while permitting them to extend their arms over
the sponge to deposit-feed. Percentage change in disk diameter of O. lineata was compared
between brittlestars living in caged and uncaged C. vaginalis (n=20). Mesh was cleaned of
epibionts every two weeks for two months.
Single sponge tubes of C. vaginalis, N. digitalis, and A. archeri were collected at Pickles
Reef from ~ 15 m depth using SCUBA in March 2005. A piece of mesh was placed over the
oscula of each tube and held in place by two small cable ties through the lip of the oscula. For
each species, sponges (n=15) were cable-tied upright to bricks on the reef and allowed to heal for
2 weeks. At the end of two weeks, O. lineata were collected from C. vaginalis, brought back to
the lab, tagged and disk diameter was measured. Within 24 hrs of collection, tagged-O. lineata
were placed inside the bottom of a meshed sponge that was removed from its healing location,
and then cable-tied upright to a brick. The sponge tubes were placed haphazardly on the reef ≥ 2
m apart at ~ 15 m depth. Meshed-sponge tubes were cleared of fouling organisms every 2 weeks
for 4 months. After 4 months, the percent change in disk diameter was calculated for all
remaining O. lineata.
Sex and Size Distribution of Ophiothrix lineata
34
The sex and size distribution of O. lineata living in multi-tubed C. vaginalis was
examined by collecting 35 individual sponges from two sites, Conch Reef, FL (n=25) and North
Dry Rocks (n=10) in May 2008. Only sponges with at least three tubes and one visible O.
lineata were collected. Sponges were cut from the base and individually placed into large 11 L
plastic bags in the field. Sponges were carefully dissected and all O. lineata were removed and
held in seawater tanks. In addition, the presence or absence of brood chambers within 30 of the
sponges was noted. Disk diameter was measured for all O. lineata and sex was determined by
visual inspection of gonads under a dissecting microscope. White testes and yellow ovaries were
easily identifiable through the tissue of larger gravid individuals; however, smaller individuals
were dissected to determine gender.
Density of O. lineata living in brooding and non-brooding C. vaginalis was compared
using analysis of covariance (ANCOVA), with total sponge surface area as the covariate.
Additional data points from surveys of C. vaginalis collected at North Dry Rocks and Pickles
Reef in which the presence of brood chambers was recorded (n=18) were added to the data of
sex distribution of O. lineata. For these added data points, O. lineata > 7 mm disk diameter were
considered mature adults based on the sex-size distribution data (Fig. 8). Both density of O.
lineata and sponge surface area were log10 transformed to meet the assumptions of ANCOVA
(Sokal and Rohlf 1981).
35
RESULTS
In total, 55, 62, and 24 choice experiments were conducted to determine the preferences of O. lineata for filtered seawater or seawater conditioned by either C. vaginalis, N. digitalis, or
A. archeri, respectively. In these experiments, O. lineata made a choice 23, 22, and 7 times, respectively. For the choice experiments comparing filtered seawater with seawater conditioned by C. vaginalis, O. lineata preferentially selected sponge-conditioned seawater (Χ2=9.78 p
<0.05). The brittlestar showed no preference between control seawater and seawater conditioned by either N. digitalis or A. archeri (Table 5). When presented with a choice between seawater
conditioned by C. vaginalis or N. digitalis, O. lineata preferentially chose the former 78.1% of
the time (p=0.001; Table 5), making a choice in 32 out of 59 assays. Based on all four choice
experiments, O. lineata was more likely to make a choice when C. vaginalis was present in the
assay than when it was not included (p = 0.0093). There were no preferences detected for assays
using crude organic extracts or frozen tissue samples of C. vaginalis (Table 6).
Disk diameter was both positively correlated with, and a good predictor of, mass of O.
2 lineata (log10 mass = 2.867·log10 disk diameter+ 0.254; R =0.941). After 4 months, growth of O.
lineata was significantly greater in long sponge tubes (12 cm) compared to short tubes (6 cm) of
C. vaginalis (Fig. 6; ANOVA F1,21= 13.0683 p = 0.0016). Average change in disk diameter of
O. lineata living in long and short sponge tubes was 2.09 mm ±0.28 SE (34% increase) and 0.79
mm ±0.20 SE (12% increase), respectively. The number of brood chambers in long sponge tubes was significantly greater than in shorter tubes (ANOVA F1,21=7.35 p=0.013; Fig. 6), with brood
chambers occurring in 5 of the short tubes and 9 of the long tubes of C. vaginalis.
There was no caging effect on the growth of O. lineata inside C. vaginalis tubes, as
growth did not differ for brittlestars living in caged C. vaginalis (6.51% ± 2.92 SE n = 17)
36
compared to uncaged C. vaginalis (3.49% ±1.96 SE n = 16; ANOVA p > 0.05) after 2 months.
After 4 months, O. lineata increased in disk diameter by 42.5%, 21.7%, and 8.8% living in C.
vaginalis, N. digitalis, and A. archeri, respectively (Fig. 7). However, loss of replicates due to
strong surge associated with hurricanes resulted in final numbers of replicates of only 2, 3 and 7,
respectively.
The proportion of gravid O. lineata increased with disk diameter, as all O. lineata < 4
mm disk diameter were juveniles and all O. lineata ≥ 8 mm were gravid (Fig. 8). Immature O.
lineata comprised 31.2% of individuals found living on C. vaginalis in May 2008. The density
of gravid O. lineata within a sponge ranged from 1-12. Of the 35 sponges surveyed, 25.7% had
either only males or only females living inside, and 28.6% had an equal ratio of males to
females. The majority of sponges (37.1%) had a greater proportion of males relative to females
(Fig. 9). An equal proportion of brooding C. vaginalis (17; 56.7%) and non-brooding sponges
(13; 43.3%) were observed (Χ2=0.53 p>0.05). There was no correlation between the presence of brood chambers and the presence of male (Χ2=0.083 p>0.05) or the presence of female O.
lineata (Χ2=0.14 p>0.05). Based on combined survey data, 18 non-brooding and 28 brooding C.
vaginalis contained at least one O. lineata. Density of both gravid and total O. lineata increased
with increasing sponge size and there was no difference in the density of either gravid or total O.
lineata between brooding and non-brooding C. vaginalis (Table 7; Fig. 10).
DISCUSSION
Host Specificity of Ophiothrix lineata
The ability of O. lineata to preferentially select the sponge C. vaginalis based on a species-specific chemical cue clearly indicates a high level of host specificity. The brittlestar
37
selected the preferred host sponge over other sponge species (Table 5) and was more active in
assays when C. vaginalis was present. The chemical cue appears to be a species-specific signal
rather than a general sponge metabolite. This adaptation to exploit a single host provides a
foundation for examining the mechanisms resulting in host specificity of O. lineata.
Host recognition using chemical cues has been described for other ophiuroid species. In
the Indo-Pacific, the juvenile brittlestar Ophiomastix venosa attaches to the disk or lies in the
bursa of adult Ophiocoma scolopendrina (a larger brittlestar) as a means to escape desiccation
during low tide events and may occasionally consume the food of the host brittlestar (Fourgon et
al. 2007; Fourgon et al. 2006; Hendler et al. 1999). Both free living and symbiotic Ophiomastix
venosa did not respond to cues from conspecifics, and only symbiotic Ophiomastix venosa
responded to chemical cues of Ophiocoma scolopendrina (Fourgon et al. 2007). The lack of
response of free-living Ophiomastix venosa to Ophiocoma scolopendrina suggests that only
Ophiomastix venosa that were previously associated with Ophiocoma scolopendrina are able to
detect the host. In the Caribbean, Ophiactis savignyi is also a facultative commensal that lives in
association with many sponge species (Clavico et al. 2006; McGovern 2002), as well as algae, gorgonians, and corals (Hendler et al. 1995). Clavico et al. (2006) found that Ophiactis savignyi preferentially migrated to sponge mimics comprised of a gel with a natural concentration of crude organic extract from the sponge Geodia corticostylifera, compared to control mimics.
While this response suggests host recognition, it is unclear if Ophiactis savignyi responded to a host-specific cue or any number of extracted primary and secondary metabolites, as G. corticostylifera was the only host examined in that study. The chemical cue used by Ophiothrix lineata appears to be a species specific signal; however, attempts to isolate the chemical signal used to detect C. vaginalis were unsuccessful. The extraction method sequestered all organic
38
metabolites from within sponge cells, many of which are intracellular and likely never
encountered by Ophiothrix lineata. Freezing and thawing would rupture sponge cells, resulting
in the leakage of cell metabolites. High concentrations of intracellular metabolites may mask or
dilute the cue used by Ophiothrix lineata for chemotaxis toward its preferred host sponge.
Host specificity can be a function of limited dispersal to new hosts (Tompkins and
Clayton 1999), and not necessarily an adaptive response to increased fitness from the association
(Timms and Read 1999). Studies of emigration and immigration indicate that large (> 5 mm
disk diameter} O. lineata remain in the same C. vaginalis for several months, but smaller O.
lineata move to unoccupied C. vaginalis (Henkel and Pawlik 2005). A recent study of gene flow among O. lineata populations in the Florida Keys suggests a southerly migration pattern that is contrary to prevailing currents and expected movement patterns based on potential larval dispersal (Richards et al. 2007). The southward direction of gene flow for O. lineata may be explained by upstream movement of small O. lineata responding to chemical cues and the ability of the brittlestar to use other sponges as temporary habitat. Ophiothrix lineata is a broadcast spawning brittlestar that does not have a larval stage, but undergoes direct development into a juvenile within 6-8 days of fertilization (Richards et al. 2007). Even with limited larval dispersal, considering the migration patterns of small O. lineata (Henkel and Pawlik 2005), there is ample opportunity for O. lineata to encounter other host sponges. The high level of specificity exhibited by O. lineata for C. vaginalis appears to be a consequence of enhanced fitness for the brittlestar from the association.
The association with C. vaginalis provides O. lineata a predation refuge as well as a place to feed (Hendler 1984; Henkel and Pawlik 2005). Refuge quality is partially dictated by the oscular diameter of sponges, as O. lineata is quickly consumed by fish predators when placed
39
in sponges like N. digitalis that have wide osculae (Henkel and Pawlik 2005). By placing a
screen over the osculae of sponges, I removed the predation limitation and were able to examine
growth of O. lineata in different sponges. While not statistically compared because of low
replication, growth of O. lineata was similar in both N. digitalis and C. vaginalis, but much less in the tube sponge A. archeri (Fig. 7). Growth rates of O. lineata living in C. vaginalis were comparable to other growth experiments for similar time periods (Fig. 6). Unlike both C. vaginalis and N. digitalis, A. archeri contains secondary metabolites that deter fish predation
(Pawlik et al. 1995) and therefore should provide a better refuge habitat. However, unlike amphipods that associate with chemically defended algae for refuge and food (Duffy and Hay
1994), O. lineata does not preferentially associate with a chemically defended refuge (Henkel and Pawlik 2005). In addition to anti-predatory effects, crude organic extracts of A. archeri have
anti-microbial properties (Kelly et al. 2003; Newbold et al. 1999). The presence of these
metabolites may make A. archeri a less hospitable habitat for O. lineata. Anti-microbial
properties may reduce development of bacterial biofilm on the sponge surface, also reducing the
nutritional quality of surface constituents compared to either C. vaginalis or N. digitalis. In
addition, A. archeri does not regularly produce larvae (Tsurumi and Reiswig 1997) on which the
brittlestars could feed. The food resources available from A. archeri are most likely less than
those offered by either N. digitalis or C. vaginalis, both of which lack chemical defenses.
Refuge quality plays some role in determining where O. lineata can survive, but food resources,
both in terms of deposit-feeding and larval predation, are more likely to impact host selection by
O. lineata. Non-preferred habitats such as A. archeri may serve as temporary hosts that provide
protection for O. lineata in search of C. vaginalis.
40
Sponge larvae provide a potential food resource to O. lineata, but O. lineata do not live
solely in brooding sponges. Of the 46 C. vaginalis with associated O. lineata surveyed in this
study, the relative number of both brooding (60.9%) and non-brooding (39.1%) sponges was not
dramatically skewed. It might be expected that the added food resource provided by sponge larvae could result in a greater number of gravid brittlestars. However, while the abundance of
O. lineata increased with sponge size, there was no difference in relative abundance of gravid O.
lineata between brooding and non-brooding C. vaginalis (Fig. 10; Table 7) and both males and
females were found proportionally in both host types. Therefore, while larval predation can
result in increased growth of O. lineata (Chapt. 1), the brittlestar does not preferentially associate
with brooding C. vaginalis.
In addition to the role of food resources in structuring the association of O. lineata with
C. vaginalis, increased probability of finding a mate could be a strong mechanism for
maintaining host specificity. The importance of mating success in structuring species-specific
associations has received little attention in marine systems (Sotka 2005); however, it is known to
be an important mechanism in insects with low mobility (Bush and Smith 1998; Hawthorne and
Via 2001). Mate localization has also been suggested to drive selection for maintaining host
specificity in four species of bat fly that have high dispersal potential to multiple hosts (Dick and
Patterson 2007). Fertilization success may be important in structuring the O. lineata-C. vaginalis
association given the low mobility of larger O. lineata. Fertilization success has been found to
increase with greater densities of free-spawning sea urchins (Levitan 2005). The abundance of
mature O. lineata increased with increasing sponge size (Fig. 10) and the tubes of C. vaginalis
are usually connected at the base, allowing movement of brittlestars between tubes within the
larger sponge. The greater density of males relative to females within a sponge (Fig. 9) may
41
increase fertilization success considering that gametes are released into a high-flow environment:
the osculum of C. vaginalis can pump 3.5 L s-1 kg-1 dry tissue (Weisz et al. 2008). The high
pumping rate of C. vaginalis may provide a dispersal mechanism for fertilized eggs that are
slightly heavier than seawater (Henkel, pers. obs.). While 65.7% of C. vaginalis surveyed had at least a 1:1 male to female ratio; a quarter had only one sex present (Fig. 9). Given the limited mobility of large O. lineata (Henkel and Pawlik 2005), solitary individuals would likely be non- reproductive, but the mobility of smaller O. lineata (Henkel and Pawlik 2005) combined with their chemotactic ability to locate host sponges would enhance pairing and reproductive success.
As noted by Hendler (1984), the deposit-feeding O. lineata is distinct among its congeners, all of which are suspension feeders. In laboratory assays, O. lineata captured swimming larvae in a fashion resembling suspension feeding, using tube feet to pass and guide larvae into its mouth (Chapt. 1). This raises some interesting evolutionary questions regarding the potential importance of both food and mating success in structuring the association of O. lineata with C. vaginalis. The increased probability of finding a mate may have resulted in a regular association with C. vaginalis, and subsequently O. lineata shifted towards deposit-
feeding on the surface of the sponge. Conversely, O. lineata may first have diverged from its
congeners as a deposit feeder, benefiting from the resources available on the surface of C.
vaginalis, as well as from larval predation. Increased mating success associated with multiple
individual O. lineata occupying a single sponge could have served to strengthen the association,
resulting in the obligate relationship the brittlestar exhibits today. Thus, considering O. lineata is unique within its genus with respect to the high level of host specificity, mode of feeding, and a direct-developing larval stage, the system provides an excellent model for examining evolutionary trade-offs and factors maintaining host specialization.
42
Host Limitations and the Distribution of Ophiothrix lineata
In the northern Florida Keys, the abundance of O. lineata is positively correlated with the
abundance of C. vaginalis on the reef (Henkel and Pawlik 2005). Furthermore, the number of large O. lineata in a single C. vaginalis rarely exceeds one individual per sponge tube, which may be a response to available food resources (Henkel and Pawlik 2005). In the present study,
O. lineata grew significantly more when living in longer tubes of C. vaginalis (Fig. 6).
Enhanced growth in longer tubes could be the result of decreased non-lethal predation (arm grazing by fish predators) on O. lineata in larger sponges, a greater surface area for deposit- feeding, or increased larval output of larger sponges.
Larger brittlestars living in small refuges may sustain increased non-lethal predation of arms, resulting in less growth of disk diameter, as more energy is diverted to healing. Arm regeneration results in reduced lipid and gonad production in the brittlestar Ophiocoma echinata
(Pomory and Lawrence 1999). Large O. lineata are cryptic during the day and only extend their arms out of sponges at night to feed when predators are less abundant (Hendler 1984). While a shorter sponge tube may increase exposure of long arms to damage, brittlestars are capable of retracting their arms into small areas for protection. In the present study, there was no observable difference in arm damage for brittlestars that had been transplanted to short and long sponge tubes. Therefore, the impact of predation is unlikely to explain the differences in growth of O. lineata in sponges of different sizes.
Another explanation for enhanced growth of O. lineata in longer sponge tubes is that longer sponges have more surface area for deposit-feeding by the brittlestar. The average arm length of O. lineata is about 10 times that of the disk diameter (Hendler 2005), hence, the O. lineata used in this study had an initial arm length of 55-80 mm, surpassing the height of the
43
short sponge tubes (60 mm). Brittlestars in long sponge tubes (120 mm) had a larger surface area for deposit-feeding, but several observations suggest this extra surface area may not be used
by O. lineata. First, as O. lineata extends its arms out of a sponge tube and over the surface to
deposit-feed, the disk does not leave the inside of the sponge and only part of the arm or arms
move over the sponge to deposit-feed (Henkel, pers. obs.). Second, O. lineata transferred to the
non-preferred vase-shaped sponge N. digitalis grew as much as individual brittlestars living in C.
vaginalis (Fig. 7), even though the surface area of N. digitalis is almost twice that of C. vaginalis
of similar height (Henkel and Pawlik 2005). This suggests that O. lineata may not differentially
benefit from a greater surface area for deposit-feeding. While deposit-feeding is the purported
primary feeding mode of O. lineata (Hendler 1984), the brittlestar is capable of consuming
larvae of C. vaginalis and N. digitalis (Henkel in prep). Increased growth of O. lineata in longer
sponge tubes may be due to the increased reproductive output of larger sponges. Long C.
vaginalis had significantly more brood chambers per tube than shorter sponge tubes (Fig. 6), and
may provide a greater larval food source for O. lineata. This difference could be due to only
41% of short sponge tubes had at least one brood chamber compared to 81% of long tubes that
had brood chambers. In summary, enhanced growth of O. lineata in longer sponge tubes is most
likely a combination of all three factors: greater shelter, more surface area for deposit-feeding,
and more opportunity to feed on larvae produced by the sponge.
The genus Ophiothrix represents a morphologically and ecologically diverse group, and
while conspecifics do not exhibit host specificity observed in O. lineata, habitat degradation has
been suggested as a cause for decreasing populations of Ophiothrix species (Hendler 2005). The
multi-tubed sponge C. vaginalis provides O. lineata with food resources and the morphology of
the sponge provides habitat for multiple gravid O. lineata within a small area. Given the host
44
specificity exhibited by O. lineata for C. vaginalis, any large reduction in C. vaginalis abundance, such as occurs from large storm events when sponges are ripped off the reef (Henkel pers. obs.), may result in decreases of O. lineata. Assuming larger O. lineata have increased
fecundity, as has been demonstrated for other brittlestar species (McGovern 2002), the
reproductive effort of O. lineata may also be limited by the size of C. vaginalis as growth of O.
lineata is reduced in smaller sponges. This limitation in reproductive output may be magnified
by the reduced frequency of mature females, as female O. lineata reach maturation at a larger size than males (Henkel in prep). In addition to growth restrictions, reduction of sponge size would reduce the abundance of O. lineata within these smaller sponges, thereby futher reducing fertilization success. The O. lineata-C. vaginalis association therefore provides a marine model for understanding the mechanisms structuring host specificity as well as constraints of an obligate life history.
LITERATURE CITED
Agosta SJ (2008) Fitness consequences of host use in the field: temporal variation in performance and a life history tradeoff in the moth Rothschildia lebeau (Saturniidae). Oecologia 157:69-82
Beck MW (1995) Size-specific shelter limitation in stone crabs: A test of the demographic bottleneck hypothesis. Ecology (Washington D C) 76:968-980
Bush GL, Smith JJ (1998) The genetics and ecology of sympatric speciation: A case study. Researches on Population Ecology 40:175-187
Caddy J (1986) Modeling stock-recruitment processes in Crustacea: some practical and theoretical perspectives. Can J Fish Aquat Sci 43:2330-2344
Caddy JF, Stamatopoulos C (1990) Mapping growth and mortality rates of crevice-dwelling organisms onto a perforated surface the relevance of cover to the carrying capacity of natural and artificial habitats. Estuarine Coastal and Shelf Science 31:87-106
45
Clavico EEG, Muricy G, da Gama BAP, Batista D, Ventura CRR, Pereira RC (2006) Ecological roles of natural products from the marine sponge Geodia corticostylifera. Marine Biology 148:479-488
Combes C (1991) Evolution of parasite life cycles. In: Toft C, Aeschlimann A, Bolis L (eds) Parasite-Host Associations: Coexistence or Conflict? Oxford University Press, Oxford, pp 66-82
Dick CW, Patterson BD (2007) Against all odds: Explaining high host specificity in dispersal- prone parasites. International Journal for Parasitology 37:871-876
Duffy JE (1992) Host use patterns and demography in a guild of tropical sponge-dwelling shrimps. Marine Ecology Progress Series 90:127-138
Duffy JE (1996a) Resource-associated population subdivision in a symbiotic coral-reef shrimp. Evolution 50:360-373
Duffy JE (1996b) Species boundaries, specialization, and the radiation of sponge-dwelling alpheid shrimp. Biological Journal of the Linnean Society 58:307-324
Duffy JE, Hay ME (1994) Herbivore resistance to seaweed chemical defense: The roles of mobility and predation risk. Ecology (Tempe) 75:1304-1319
Engel S, Pawlik JR (2000) Allelopathic activities of sponge extracts. Marine Ecology Progress Series 207:273-281
Faucci A, Toonen RJ, Hadfield MG (2007) Host shift and speciation in a coral-feeding nudibranch. Proceedings of the Royal Society B-Biological Sciences 274:111-119
Fourgon D, Jangoux M, Eeckhaut I (2007) Biology of a "babysitting" symbiosis in brittle stars: analysis of the interactions between Ophiomastix venosa and Ophiocoma scolopendrina. Invertebrate Biology 126:385-395
Fourgon D, Lepoint G, Eeckhaut I (2006) Assessment of trophic relationships between symbiotic tropical ophiuroids using C and N stable isotope analysis. Journal of the Marine Biological Association of the United Kingdom 86:1443-1447
Giorgi MS et al. (2004) Causal mechanisms underlying host specificity in bat ectoparasites. Oecologia 138:648-654
Hawthorne DJ, Via S (2001) Genetic linkage of ecological specialization and reproductive isolation in pea aphids. Nature (London) 412:904-907
Hay ME, Duffy JE, Pfister CA, Fenical W (1987) Chemical Defense against Different Marine Herbivores Are Amphipods Insect Equivalents? Ecology 68:1567-1580
Hendler G (1984) The association of Ophiothrix lineata and Callyspongia vaginalis: a brittlestar- sponge cleaning symbiosis? Marine Ecology 5:9-28
46
Hendler G (2005) Two new brittle star species of the genus Ophiothrix (Echinodermata:Ophiuroidea:Ophiotrichidae) from coral reefs in the Southern Caribbean sea, with notes on their biology. Caribbean Journal of Science 41:583-599
Hendler G, Grygier MJ, Maldonado E, Denton J (1999) Babysitting brittle stars: Heterospecific symbiosis between ophiuroids (Echinodermata). Invertebrate Biology 118:190-201
Hendler G, Miller JE, Pawson DL, Porter MK (1995) Sea stars, sea urchins, and allies: echinoderms of Florida and the Caribbean. Smithsonian Institution Press, Washington, DC
Henkel TP, Pawlik JR (2005) Habitat use by sponge-dwelling brittlestars. Marine Biology 146:301-313
Hufbauer RA, Via S (1999) Evolution of an aphid-parasitoid interaction: Variation in resistance to parasitism among aphid populations specialized on different plants. Evolution 53:1435-1445
Johnston L, Miller MW (2007) Variation in life-history traits of the corallivorous gastropod Coralliophila abbreviata on three coral hosts. Marine Biology 150:1215-1225
Kelly SR, Jensen PR, Henkel TP, Fenical W, Pawlik JR (2003) Effects of Caribbean sponge extracts on bacterial attachment. Aquatic Microbial Ecology 31:175-182
Kissling DL, Taylor GT (1977) Habitat factors for reef-dwelling ophiuroids in the Florida Keys. Proceedings Int. Coral Reef Symp. 3:225-231
Levitan DR (2005) The distribution of male and female reproductive success in a broadcast spawning marine invertebrate. Integrative and Comparative Biology 45:848-855
Macdonald KS, Rios R, Duffy JE (2006) Biodiversity, host specificity, and dominance by eusocial species among sponge-dwelling alpheid shrimp on the Belize Barrier Reef. Diversity and Distributions 12:165-178
McGovern TM (2002) Patterns of sexual and asexual reproduction in the brittle star Ophiactis savignyi in the Florida Keys. Marine Ecology Progress Series 230:119-126
Newbold RW, Jensen PR, Fenical W, Pawlik JR (1999) Antimicrobial activity of Caribbean sponge extracts. Aquatic Microbial Ecology 19:279-284
Pawlik JR, Chanas B, Toonen RJ, Fenical W (1995) Defenses of Caribbean sponges against predatory reef fish. I. Chemical deterrency. Marine Ecology Progress Series 127:183-194
Pomory CM, Lawrence JM (1999) Effect of arm regeneration on energy storage and gonad production in Ophiocoma echinata (Echinodermata: Ophiuroidea). Marine Biology 135:57-63
47
Richards VP, Thomas JD, Stanhope MJ, Shivji MS (2007) Genetic connectivity in the Florida reef system: comparative phylogeography of commensal invertebrates with contrasting reproductive strategies. Molecular Ecology 16:139-157
Sloan N, Campbell A (1982) Perception of food. In: Jangoux M, Lawrence J (eds) Echindoerm Nutrition. CRC Press, Rotterdam, pp 3-23
Sokal RR, Rohlf FJ (1981) Biometry: the principles and practice of statistics in biological research, 2d edn. W. H. Freeman, San Francisco
Sotka EE (2005) Local adaptation in host use among marine invertebrates. Ecology Letters 8:448-459
Sotka EE (2007) Restricted host use by the herbivorous amphipod Peramphithoe tea is motivated by food quality and abiotic refuge. Marine Biology 151:1831-1838
Timms R, Read AF (1999) What makes a specialist special? Trends in Ecology and Evolution 14:333-334
Tompkins DM, Clayton DH (1999) Host resources govern the specificity of swiftlet lice: Size matters. Journal of Animal Ecology 68:489-500
Tsurumi M, Reiswig HM (1997) Sexual versus asexual reproduction in an oviparous rope-form sponge, Aplysina cauliformis (Porifera; Verongida). Invertebrate Reproduction and Development 32:1-9
Weisz JB, Lindquist N, Martens CS (2008) Do associated microbial abundances impact marine demosponge pumping rates and tissue densities? Oecologia 155:367-376
Wulff JL (2006) Ecological interactions of marine sponges. Canadian Journal of Zoology 84:146-166
48
Table 5: Results of paired chemical cue choice assays in which O. lineata was presented sponge-conditioned seawater and filtered seawater, or seawater conditioned with C. vaginalis and N. digitalis. The percentage of choices for each treatment is given relative to the number of choices made (n) out of the total number of assays run. Asterisk denotes a significant difference from an expected equal probability of selecting either treatment using chi-squared goodness of fit.
Paired Percentage Total Number Treatments of Choices n of Assays Sponge Seawater : Control Seawater C. vaginalis 82.6%* 23 55 N. digitalis 50.0% 22 62 A. archeri 42.9% 7 31
C. vaginalis : N. digitalis C. vaginalis 78.1%* 32 59
49
Table 6: Choice experiments using chemical cues isolated from C. vaginalis and presented to brittlestars using different techniques and at various concentrations.
Number of Choices Seawater Treatment (ml) Soak Time Sponge Control No Choice Carragenan Gels 80 ml gel at natural concentration 5 L 4 hrs 0 4 1 40 ml gel at natural concentration 5 L 4 hrs 0 3 0 20 ml gel at natural concentration 5 L 4 hrs 0 3 6 20 ml gel at 50% dilution 5 L 20 hrs 0 0 6 50 ml gel at 20% dilution 5 L 4 hrs 0 1 7
Frozen sponge 5-10 ml tissue 1-2 L 1 sec -5 min 0-4 0-4 0-3 (n=33)
10 ml Crude Extract Resuspended in 10 ml Methanol 10-50 drops (n=12 trials) 1 L - 0 0-2 1-5
50
Table 7: Results from linear regression analyses and ANCOVA for number of gravid (a) and total (b) O. lineata found living in brooding (n=28) and non-brooding (n=18) C. vaginalis.
Regression ANCOVA Group Equation R2 Source dfMS F P a) log (gravid O. lineata + 1) (Y) vs. log sponge surface area (X) Brooding log Y = 0.853 X - 1.576 0.735*** Treatment 1 0.018 0.64 0.4278 Non-Brooding log Y = 0.700 X - 1.263 0.686*** Slope 1 0.028 1.00 0.3242 Covariate 1 2.880 103.09 <0.0001 b) log total O. lineata (Y) vs. log sponge surface area (X) Brooding log Y = 0.860 X - 1.495 0.610*** Treatment 1 0.023 0.49 0.4862 Non-Brooding log Y = 0.701 X - 1.143 0.608*** Slope 1 0.030 0.66 0.4201 Covariate 1 2.909 63.63 <0.0001 *** P < 0.0001
51
40% 18 MEAN NUMBER OF BROOD CHAMBERS PER TUBE BROOD CHAMBERS PER MEAN NUMBER OF 16
14 30%
12
10 20% 8
6
10% 4 PERCENT CHANGE IN DISK DIAMETER CHANGE PERCENT
2
0% 0 Long Tubes Short Tubes
Fig. 6: After 4 months, percentage change in disk diameter of O. lineata (bars ±SE) was significantly greater for O. lineata living in long (n=11) and short (n=12) C. vaginalis sponge tubes (ANOVA F1,21= 13.0683 p = 0.0016). Long tubes had significantly more brood chambers per tube (squares ±SE) than short tubes (ANOVA F1,21=7.35 p=0.013).
52
100%
80%
60%
40%
20% PERCENT CHANGE IN DISK DIAMETER CHANGE PERCENT
0% C. vaginalis N. digitalis A. archeri
Fig. 7: Mean percentage change in disk diameter of O. lineata (±SE) caged inside three sponge species for 4 months: Callyspongia vaginalis (n=2), Niphates digitalis (n=7), and Aplysina archeri (n=3). Despite low replication, mean growth of O. lineata in C. vaginalis was similar to that from previous experiments (see Fig. 6).
53
25% 100%
20% 80% PERCENT MATURE PERCENT
15% 60%
10% 40% FREQUENCY
5% 20%
0% 0% 23456789101112 DISK DIAMETER (mm)
Fig. 8: Size frequency distribution (bars) and the percent of mature brittlestars at each size class (diamonds) of the 204 O. lineata living inside 35 C. vaginalis.
54
35%
30%
25%
20%
15% FREQUENCY
10%
5%
0% All 0.30.511.523.54 5 All Females Males
RATIO OF MALE TO FEMALE O. lineata PER SPONGE
Fig. 9: Frequency of observed ratios of male to female O. lineata living in a single C. vaginalis (n=35).
55
1.4 a) 1.2 +1) 1
lineata 0.8
O.
0.6 (gravid
10 0.4 log
0.2
0 1.5 2 2.5 3 3.5
b) 1.4
1.2
1 lineata 0.8 O.
0.6 total 10
log 0.4
0.2
0 1.5 2 2.5 3 3.5 log10 sponge surface area
Fig. 10: Abundance of gravid Ophiothrix lineata (a) and total abundance of O. lineata (b) as a function of surface area of C. vaginalis. Sponges with brood chambers (closed circle, solid line; n=28) and without brood chambers (open circle, dashed line; n=18). Regression and ANCOVA analysis in Table 7.
56
CHAPTER 3: LIFE HISTORY TRAITS OF THE SPONGE-DWELLING BRITTLESTAR OPHIOTHRIX LINEATA
57
ABSTRACT
Life history strategies are the product of multiple variables, including mode of
reproduction, growth rates of larvae, juveniles and adults, and time to reproductive maturation. I
examined development and growth in the obligate sponge-dwelling Caribbean brittlestar
Ophiothrix lineata, which lives in association with the tube sponge Callyspongia vaginalis. In
the laboratory, spawning of O. lineata was induced by sudden exposure to bright light, with
females releasing eggs within 5 minutes of males releasing sperm. Fertilized eggs averaged 330
μm in diameter and embryos within the fertilization membrane averaged 230 μm. Over 6 days,
O. lineata underwent complete development within a firm, adhesive fertilization membrane, after
which juveniles ~ 150 μm disk diameter hatched. This is the first report of non-pelagic direct
development in the genus Ophiothrix; but morphological similarities suggest that this
developmental mode may be common to other species. Mean disk diameter of gravid O. lineata
was 8.6 mm. Size at sexual maturity differed slightly between male and female O. lineata, as males ranged in size from 4.8-12.5 mm while females had disk diameter of 6.1-12.3 mm.
Growth of O. lineata was measured using a range of initial disk diameters, and growth models were used to calculate size-at-age estimates. Average specific growth rate was 3.2 mm yr-1± 4.9
SD, and decreased significantly with increasing disk diameter. Growth rates were highest for brittlestars 5-8.5 mm disk diameter, coinciding with average size of O. lineata. Growth models yielded estimates of 1 year of growth as 5-8.3 mm disk diameter. Age estimates of the largest O. lineata surveyed, 12.5 mm diameter, varied between 4.4 and 8.7 years based on the determinate growth model and the indeterminate Tanaka model, respectively. Decreased growth rates in larger O. lineata may be a function of increased energy expenditure on reproduction.
58
INTRODUCTION
Comparisons of life histories of closely related species with different strategies of growth and reproduction can provide insight into the evolution of observed traits (Roff 1992). Studies of
echinoderms have provided much of the theoretical framework for understanding the evolution
of larval development; however, most of the focus has been on asteroids and echinoids
(McEdward and Janies 1993; Strathmann 1975; Strathmann 1978). Larval development has
been described in less than 5% of the over 2,000 extant species of ophiuroids; therefore, the lack
of information prevents evolutionary comparisons within this diverse group (McEdward and
Miner 2001).
Marine invertebrates often have a larval dispersal stage, which is typically very different
from the adult in form and feeding mode. Larval development can be characterized by the type
of morphogenesis, simple or complex; nutritional source, either planktotrophic or lecithotrophic;
and location of development, pelagic or benthic (McEdward and Janies 1997). The presence of a
distinct larval stage that metamorphoses into an adult differs from direct development, in which a juvenile hatches from the egg (Emlet 2006).
Ophiuroids generally have three main larval types: the planktonic ophiopluteus that can
be both feeding and non-feeding, a non-feeding vitellariae, and direct development (Hendler
1975; Hendler 1991; McEdward and Miner 2001). The family Ophiothrichadae includes a
cosmopolitan group of species with various life history strategies, such as the genus
Macrophiothrix, which contains species having obligate and facultative feeding larval forms as
well as lecithotrophic larvae (Allen and Podolsky 2007). Also within the family
Ophiothrichadae is the genus Ophiothrix, from which the larval development of 8 species has
been described. Seven of these 8 species have a pelagic planktotrophic 8-arm ophiopluteus
59
larval stage (McEdward and Miner 2001; Selvakumaraswamy and Byrne 2000). Only,
Ophiothrix orstedii produces a lecithotrophic 2-arm ophiopluteous (Mladenov 1979).
Additionally, a recently described species, Ophiothrix stri, produces large demersal embryos within a spacious fertilization membrane (Hendler 2005), which is indicative of abbreviated lecithotrophic development (Hendler 1991), although the full development of this species has yet
to be described.
While species of Ophiothrix are often associated with sessile invertebrates, algae, or
occur in large intraspecific aggregations (Hendler 2005; Hendler et al. 1995; Morgan and
Jangoux 2005), Ophiothrix lineata lives in a species-specific association with the tube sponge
Callyspongia vaginalis (Hendler 1984; Henkel and Pawlik 2005). Adult O. lineata exhibit
strong specificity for C. vaginalis and are able to detect the preferred sponge using chemical cues
(Henkel Chapt. 2). The sponge provides refuge and food resources for the brittlestar, as O.
lineata lives inside the tubes of C. vaginalis, deposit-feeding on the sponge surface, and likely
consuming sponge larvae (Hendler 1984; Henkel and Pawlik 2005; Henkel Chapt. 1). Multiple
O. lineata regularly occupy a single sponge, and the presence of both male and female
brittlestars in a single sponge may increase the fertilization success of O. lineata (Henkel Chapt.
2).
In the present study, I describe the development of O. lineata from post-fertilization to
juvenile, providing comparisons with other species in the genus Ophiothrix. I also present
morphometric data collected from multiple surveys examining the association of O. lineata with
C. vaginalis. Finally, I examine the growth of O. lineata living in association with C. vaginalis,
and use multiple growth models to develop size-at-age estimates to explore relationships
between growth and reproduction in O. lineata.
60
METHODS
Larval Development
Eleven O. lineata living in association with C. vaginalis were collected from a shallow reef off Palm Beach, Florida (26o 43.077 N 80o 01.774 W). Brittlestars were transported in a 5 gal bucket with seawater to the lab and maintained in seawater at 24 oC. Light shock was used to induce spawning by placing brittlestars directly under a light for 30 minutes, followed by 130 minutes of darkness. Brittlestars were exposed to the light a second time and began spawning after 5 minutes. Gametes were allowed to mix for ~1 hr and eggs were transferred to fresh seawater in glass bowls. Development of embryos was monitored using light microscopy, and eggs were maintained with daily changes of seawater. Seawater was lightly agitated using an aquarium bubbler and average water temperature was 23 oC.
Morphometrics and Size Distributions
Morphometric data for O. lineata, including disk diameter, longest arm length, and mass, were compiled from multiple collections of O. lineata living in association with C. vaginalis from 2005-2008 on shallow reefs off the northern Florida Keys [North Dry Rocks (25o 07.850 N;
80o 17.521W), Pickles Reef (24 o 59.286 N; 80o24.6 W), Conch Reef (24o 57.01 N; 80o 27.25
W)]. Disk diameter, measured as the distance from the base of one arm to the point on the disk directly opposite, was measured using calipers. Longest arm length was determined using a ruler to the nearest mm and wet mass was measured with an analytical balance. Brittlestars were patted with a paper towel prior to measurements to remove excess seawater. Initial morphometric data were taken, then individual O. lineata were placed in pre-weighed tins and dried in an oven at 60 oC for 36-48 hours, or until no change in mass was detected. Dry mass
61
was correlated to wet mass for 67 O. lineata using linear regression. The resulting regression
was used to calculate dry mass for all O. lineata. Calculated dry mass (n=28) and longest arm
(n=239) were correlated to disk diameter using linear regression of log10-transformed data.
In May 2008, the sex of 165 O. lineata was determined by visual inspection of gonads under a dissecting microscope for brittlestars collected from North Dry Rocks and Conch Reef.
White testes and yellow ovaries were easily identifiable through the tissue of larger gravid individuals, while smaller individuals were dissected to observe gonads. Disk diameter was compared between male and female O. lineata using nonparametric Wilcoxon rank sums. Wet mass and length of longest arm were compared between male and female O. lineata using analysis of covariance on log transformed data. Wet mass and longest arm length were the response variables and disk diameter was used as the covariate. All statistics were performed in
JMP 7.0 (SAS Institute).
Growth Model
Growth experiments were conducted on a shallow patch reef, 10 m in depth, at North Dry
Rocks. Single tubes of C. vaginalis, ~12 cm in height, were cut using a scalpel and cable-tied upright to bricks that were secured to coral pavement and placed haphazardly on the reef.
Sponges were allowed to heal for seven days. At the end of two weeks, O. lineata were collected from C. vaginalis, brought back to the lab, and their oral surfaces were tagged with small dots of the histological dye Congo Red. Tagged brittlestars were kept in recirculating seawater overnight to confirm survival and quality of tags. The following day, initial disk diameter was measured to the nearest 0.5 mm using calipers, and brittlestars were taken to the field. A single O. lineata was placed in each tube of C. vaginalis. At the end of the experiment,
62
brittlestars were recollected, brought back to the lab, and final disk diameter was measured.
Specific growth rate was calculated as: