Effects of Removing Sea Urchins (Strongylocentrotus ...

2 downloads 0 Views 2MB Size Report
Sep 20, 1995 - Author(s): Hans Petter Leinaas and Hartvig Christie. Reviewed work(s): ...... Dayton PK, Tegner MJ, Parnell PE, Edwards PB (1992) Temporal.
International Association for Ecology

Effects of Removing Sea Urchins (Strongylocentrotus droebachiensis): Stability of the Barren State and Succession of Kelp Forest Recovery in the East Atlantic Author(s): Hans Petter Leinaas and Hartvig Christie Reviewed work(s): Source: Oecologia, Vol. 105, No. 4 (1996), pp. 524-536 Published by: Springer in cooperation with International Association for Ecology Stable URL: http://www.jstor.org/stable/4221217 . Accessed: 04/10/2012 13:25 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and extend access to Oecologia.

http://www.jstor.org

Oecologia (1996) 105: 524-536

Petter

Effects stability of

kelp

Leinaas of

?

removing

of

the

forest

Christie

Hartvig

sea

urchins

barren

state

recovery

in

Received 22 March 1995 / Accepted:

(Strongylocentrotus

and the

droebachiensis):

succession

east

Atlantic

20 September 1995

state of of the barren properties in studied were urchin system the sea urchin The of Norway. ability to maintain droebachiensis high Strongylocentrotus and recover from perturbations, densities population were of kelp forest revegetation, and the succession studied experimentally by reducing the sea urchin denwas information sity on a barren skerry. Additional a obtained from community natural, changes following that varied between but patchy, sea urchin mortality sites. On the barren grounds, high sea urchin densities by annual recruitment. (30-50 per m2) is maintained initiated luxof sea urchin densities Severe reductions reductions while more moderate uriant kelp growth, of opportunistic establishment allowed algae (during of but no kelps. Succession spring and early summer), in urchin dendecline sea after the severe algal growth, a predictable sities, followed pattern. At first the substrate was colonized algae, but within by filamentous few weeks they were outcompeted by the fast growing 3-4 years of the saccharina. After kelp Laminaria the slower-growing, removal long-lived experiment, dominant. increasingly kelp L. hyp er borea became in sea urchin after reduction Increased food availability of the to increased individual led growth density denthe population sea urchins. However, remaining nor neither from recruitment sity did not increase, from adjacent areas with high sea urchin immigration of a dense kelp densities. Possibly, early establishment in a the stand, may represent ability of sea breakpoint

Abstract a kelp northern

1996

PAPER

ORIGINAL

Hans

Springer-Verlag

Stability forest-sea

H. P. Leinaas (^)! ? H. Christie Norwegian Institute for Nature Research, P.O. Box 1037, Blindem N-0315 Oslo, Norway Present address: university of Oslo, Department of Biology, Division of Zoology, P.O. Box 1050, Blindem, N-0316 Oslo, Norway, Fax: + 47 22 85 4605

a barren state. The ability of to reestablish urchins an area L. saccharina quickly to invade and monopolize and effects on the suchave both negative positive may forest. the climax L. hyperborea cession towards kelp but interactions may slow the process, Competitive L. will also stand of saccharina of a dense development recruits of the more reduce grazing risk on scattered slowly growing L. hyperborea. Key words Kelp ? Stability ground

- sea urchin ? Succession

interactions

? Barren

Introduction of how is a complex function stability Community species interact with each other (e.g. Pimm 1986). Some in structuring a are particularly dominant interactions to study in and correspondingly important community, the persisthe forces underlying order to understand contence and changes of the system. Central questions cerning stability are the degree to which a system may or being transferred without be perturbed collapsing to a different state, how fast it returns to the original and whether there exist a perturbation, state following several stable states of the system (May 1977; Connell and Sousa 1983; Pimm 1984). and degree of In a grazer-limited system, dynamics react to varion how the both grazers stability depend and how the prey species react ation in food availability we in grazer to variations Normally, populations. to a would expect that growth in a grazer population would create an unstable level leading to overgrazing followed situation by a decline in the grazer density. sea urchins in marine kelp forest habitats, However, densities at long after their very high may persist of algal vegetahas led to destruction initial outbreak et al. 1978; Simenstad tion (Lang and Mann 1976; Keats Johnson and Mann 1982; Dayton 1985; 1991). In this "barren ground" situation they feed on diatoms,

OECOLOGIA habirecruits, drift algae from intertidal macrophyte Vadas tats etc. (Lang and Mann 1976; Chapman 1981; Such sea urchin et al. 1986; personal observations). over may, however, suffer severe mortality populations increase in algal growth large areas with a subsequent et al. 1980; 1979; Boudouresque (Pearce and Hines Miller and Colodey 1983; Scheibling 1986). These draand matic changes between productive algal vegetation the importance of kelp-sea a barren state emphasize in the dynamics of kelp forest comurchin interactions munities. They also represent an interesting system for studying aspects of stability, like the recovery after perof breakpoints turbations and possible existence (May turns towards 1977; Pimm 1986) where the system of another and the successional state, pathways this development. We have focused on these questions in northern where the sea urchin StrongyloNorway, centrotus droebachiensis has caused kelp forest destruckilometers of coastline tions along several hundred 1982; Hagen years (Sivertsen during the last 15-20 et al. 1995). Skadsheim 1983,1987; in the North is widely distributed S. droebachiensis and the Northeast Pacific Atlantic 1974; (Jensen for and Gilkinson 1994), and is responsible Gagnon of kelps world-wide extensive (Breen and overgrazing et al. 1994). This Mann 1976; Duggins 1980; Skadsheim has been extensively studied in the northphenomenon west Atlantic, where sea urchins have caused destruction of the kelp forest along Nova Scotia and Maine 1981; (Breen and Mann 1976; Mann 1977; Chapman Harris 1981; Johnson and Mann 1982; Vadas et al. off Nova Scotia was first 1986). Kelp forest destruction 1970 (Mann 1977). Some areas have reported around remained barren for at least 15-20 years (Miller 1985; 1986; Keats 1991), but sea urchin mass morScheibling infection led to recovery of kelp due to amoebic tality over wide areas in the early 1980s (Jones forests and Scheibling 1985; Miller 1986; 1985; Scheibling and Raymond America, 1990). In North Scheibling several authors have also studied effects of removing from experimental sites (exclosure) S. droebachiensis within barren areas (Breen and Mann 1976; Duggins et al. 1983; Keats 1980; Chapman 1981; Himmelman et al. 1990). Laminaria The dominant hyperkelp in Norway, and long lived (15-20 borea, is slow-growing years) to kelps exposed to grazing by (Kain 1971), compared in North America which have life S. droebachiensis 1980; Johnson and Mann spans of 2-4 years (Duggins to sea urchin removal 1988). Community responses be the same in different therefore not necessarily may A (5 years) study on these responses regions. long-term sea urchins was initiated by experimentally removing In the 3rd from a barren skerry in northern Norway. us with year of this study, a mass mortality supplied reduced sea urchin densities additional sites with (Christie et al. 1995). The general goals of this research were to study aspects of community stability of the sea-

105(1996)

? Springer-Verlag

525

urchin- dominated barren state and the transition of the system towards a kelp-forest-dominated state. we wanted to elucidate: (1) primary Specificially, of to reduced responses algae grazing pressure, and sucin the cessional involved development processes towards a L. hyperborea kelp forest, and (2) the organism and population of sea urchins to responses In of macroalgal improved availability food-plants. in the mass mortality addition, variability gave valuin sea urchin denabout the reduction able information sity needed to initiate kelp recovery. Materials

and methods

Study area The work presented here is part of a project on the dynamics and stability of kelp forest communities near Vega Island in northern Norway (Fig. 1). This region is characterized by a shallow rocky subtidal zone, with thousands of small islands and skerries, reaching out to about 50 km off the main coastline. L. hyperborea used to dominate the kelp forests of this area (Grenager 1955; Kain 1971). According to the local fishermen, the dense kelp forests persisted up to the early 1970s when the first barren areas were observed. When this project started in 1988, barren grounds inhabited by high densities (30-50 per m2) of S. droebachiensis covered most of the shallow shelf, except for an outer band, 5-15 km wide, which was dominated by luxuriant forests of L. hyperborea. The study was carried out about 15 km shorewards from the outer kelp forest, in a group of small skerries 2-3 km north of Vega (Fig. 1). In 1988 the rocky subtidal zone of the whole area was barren, while lack of sea urchins allowed a dense vegetation of fucaceans intertidally. Neither L. hyperborea nor L. saccharina was observed in this intertidal zone, which therefore apparently did not represent a potential refuge for these subtidal kelps. For experimental purposes, we chose a pair of adjacent skerries (sites 1 Fig. 1 The study sites in a group of small iselets and skerries north of Vega Island: 1 experimental skerry, 2 control, 3, 4 and 5 other sites where sea urchin density and algal cover have been recorded since 1991 (Steinskjaer, Nilsarentsskjaer and Sando respectively). Unnumbered arrows point to sites where persistence of the barren state was observed until summer 1991

VEGA

ISLAND

526

OECOLOGIA

105(1996)

? Springer-Verlag

and 2). On one of them, here called the experimental skerry, we reduced the sea urchin density, while the other served as an untreated control. At low tide, the emersed part of the experimental skerry was about 50 ? 20 m, while the control was 10 ? 10 m with a shallow subtidal ridge running eastwards for another 30-40 m. The rocky surface slopes down to a sandy bottom at 10 12 m on all sides of both skerries, except for a 5 m wide and 20 m long area between the two skerries, where the depth is 5 m. The rocky barren area was roughly estimated to represent 5-10000 m2 on both skerries, with the subtidal slope mostly covering a distance of 30-40 m.

Methods Trophic interactions and community responses to changes in species dominance, are commonly studied by caging experiments with standarized replicate units. One problem with this approach is that ecological processes are sensitive to differences in spatial scale over a range that may far exceed the sizes of the caging units (Wiens 1989, Levin 1992). It is therefore also important to study effects of manipulations at larger scales. To address this scaling problem, we took advantage of the topography in the area, with patches of sea urchin habitat (skerries) bounded by a "matrix" of non-habitat sandy bottom, which allowed manipulation of an entire habitat patch, and thereby study of community stability on a more natural scale than when using sea urchin exclosures. The easiest way to kill large numbers of sea urchins is with quicklime (Bernstein and Welsford 1982), which could have been done on several skerries. It is impossible, however, to keep the effect within well defined borders and not affect control areas or other organisms. Therefore, using SCUBA diving, we manually crushed all sea urchins observed between 0 and 10 m depth around the experimental skerry. The method was too time-consuming and costly to be carried out on more than one skerry. The experiment included two periods; a pilot project when sea urchin densities were reduced on four occasions between October 1988 and June 1989, and the main project, studying long-term effects on algae and sea urchins at this reduced density, until October 1993. Sea urchins were removed repeatedly during the pilot period to ensure that their density remained low during one recruitment period for the main algal species in the area. The removals were done by four divers on 18-19 October 1988, and two divers on 9 March, 17 April and 12 June 1989. To compensate for the lack of replicate treatments, we monitored persistence of the barren state on 11 other sites in the area once a year (see map, Fig. 1). This was done to evaluate whether vegetation changes on the experimental skerry could be due to natural fluctuations in the area independent of the treatment. Beginning in April 1991, we started collecting data on sea urchin densities and algal cover on three of these sites (Steinskja?r, Nilsarentskjaer, Sando) on the same dates as in the experimental study. We also planned a supplementary caging experiment with replicates on several skerries, which, however, was spoiled by a mass mortality of S. droebachiensis in summer 1991. This event instead represented a "natural experiment" offering an opportunity to study effects of reduced sea urchin density on five habitat patches, including our two main skerries, from which we had quantitative data before the mortality. Consequently, effects of sea urchin reductions were studied on a between-patch scale by combining the removal experiment with replicate controls and the "natural experiment". The study of succession processes, however, did not involve tests of different treatment effects and therefore could be done by replicates within the experimental skerry, without being biased by pseudoreplication. Sea urchin densities and algal cover were estimated, using SCUBA, at 2, 5 and 10 m depth along a 10-15 m wide vertical transect on the south sides of both the experimental and control skerries on 18 dates from October 1988 to October 1993. The 0 m depth was defined as the lower limit of intertidal fucaceans. Once a year, a similar transect was monitored on the western side of the experi-

mental skerry, which is more wave-exposed than the rest of the skerry. (The transect chosen initially on the control skerry ended on a sandy bottom at 6-7 m, but from July 1990 it was redirected slightly to the east to extend below 10 m in order to include data from this depth). Sea urchin densities were estimated haphazardly with ten 0.25-m2 quadrats at the three depths. Algal cover was estimated from five replicate photographs taken in a 0.25-m2 frame. Percentage cover was estimated from a grid with 100 points placed above the picture, and recording the species underneath each point. The length of our transects down to 10 m was about 50 m on the control and 30 m on the experimental skerry. The same methods were also used on the three additional sites. Non-destructive photographic registration is suitable for repeated sampling of a transect. However, tiny algae may give the same cover as large kelps, and plants shaded by larger ones are not recorded. Additional data on vegetation structure needed for the analysis were obtained by harvesting algae within replicate quadrats (0.25 m2) at 2 m depth eight times between June 1989 and October 1993. At first we only harvested the west side, but after noticing differences in kelp growth on the two sides of the skerry in autumn 1991, we started also to harvest the south side, from a 10 m wide area just east of the main transect. The number of replicates (three) was kept low to reduce disturbance. However, on the last sampling the number was increased to seven, to improve the test on difference in abundance of the two kelp species at the end of the study. The algae were analysed for total wet weight, weight of dominant species, and species composition. Wet weight was measured after dripdrainage for 10 min. For the Laminaria species, we also counted the numbers and their total length. L. hyperborea was aged by counting annual rings on cross-sections of the stipes, a method not possible for L. saccharina (Kain 1963). After the sea urchin mortality in 1991 with a subsequent algal bloom on the control, the western side of this skerry was also harvested by the same method. Size distribution of sea urchins (test diameter) was determined with callipers, to the nearest 1 mm, on four occasions between July 1990 and November 1991. Each analysis was based on 60-120 animals collected 10 50 m east of the main transect areas. Samples from the first two dates (before the population crash in 1991) were taken inside a 0.25-m2 frame placed haphazardly at 5 m until a given number of sea urchins was obtained. Later, sea urchins were collected along 1.5-m-wide and up to 20-m-long areas to obtain similar sample sizes. We did not consider sea urchins below 10 mm diameter. Due to their small size and mainly cryptic behaviour, they are not easily observed, and are not part of the population exploiting resources on the open substrate. Since 1992 few sea urchins have been found on the skerries, and no more samples for size analyses were taken. Growth rings of the sea urchins (Jensen 1969) were analysed in July 1990 and November 1991.

Analysis Differences in sea urchin density between the two skerries were tested by two-way ANOVA with fixed factors site and depth (2 and 5 m) at the start of the experiment, and at the last sampling date before the mass mortality in 1991 (i.e. October 1988 and April 1991). Differences between two dates on the same skerry due to the mass mortality were analysed by two-way ANOVA with factors time and depth. Since a strong effect of depth would usually mask effects of differences between skerries or times in the full model, we here give the partial F ratios (and ? values). Comparisons of sea urchin size frequencies between pairs of samples were tested by the Kolmogorov-Smirnoff two sample test (K-S test). Growth rates of sea urchins were analysed by regressions between test diameter and number of growth zones, and regression lines were compared by testing for differences in slope or elevation (analysis of covariance). For the ANOVA, population density data were log-transformed and test diameter values were square-root-transformed to homogenize variances.

OECOLOGIA

105(1996)

? Springer-Verlag

Experimental

Results estimates of S. droebachiensis from the experDensity imental and control skerries are shown in Fig 2. At the start there were no significant in density differences the skerries (F(L29) = 0.37, ? = 0.54). Effect of between = = 0.03), while the depth was significant (F(K29) 5.24, ? in vertical distribution between sites (redivergence at 5 m on the control) was not latively few animals sufficient to give significant cross effect of site ? depth = (^(1.29) 3.18, ? = 0.08). Later, the vertical distribution on the control became more similar to the pattern originally observed on the experimental skerry. For the next 3 years, the population density at the control remained while on the experimental high and fairly constant, was reduced from about 30 per m2 at skerry density both 2 and 5 m, to about 10 and 5-10 per m2 respecbetween skerries persisted until tively. This difference = < 1991 ? and the effects of 33.07, April (F0M) 10"5), = 9.84, ? = with depth was still significant (F{Ub) 0.003) no cross effects site ? depth (F(K36) = 1.63, ? = 0.21). Density

at 10 m was

more

variable

and

difficult

Control

527

skerry

skerry

S

to

of Stronglocentrus droeFig. 2 Mean density ?1 SE {n=\0) bachiensis estimated at 2, 5 and 10 m on the experimental (O) and control (?) skerries between October 1988 and October 1993. (The sea urchin density at 2 m depth was not recorded on the experimental skerry in April 1990 due to very dense algal cover)

8. i

1988 1989 1990 1991 1992 1993 Fig. 3 Development in wet weight (D) and population density (no. of plants per m2) (?) of the kelp Laminaria saccharina at 2 m depth on the experimental and control skerries from July 1988 to July 1993. Each value is a mean (? 1 SE) of 3 replicates (0.25 m2)

in part because data were lacking from the the first two years. By June 1991, severe morlevelled out the differences tality on both skerries between the two sites. Benthic to the treatment. algae reacted strongly While the control remained barren, a luxuriant algal soon developed on the experimental vegetation skerry. Small filamentous and a few kelp recruits algae (< 5 cm) were first observed in April 1989, but only on the western side where most of the they covered substrate. By June, a dense algal vegetation had develall around the skerry oped on the upper subtidal (Table 1), with L. saccharina constituting two-thirds of the harvested biomass. For the next 2-3 years, this fastgrowing kelp made up about 95% of the algal biomass (Table 1). in biomass and population Changes density (number of plants per m2) of L. saccharina at 2 m on the west side of the skerry, are shown in Fig. 3. Strong in spring 1989, resulted in several hundred recruitment

analyze, control e w J3 $> O 8?? e ^

S

?I

li ? ?a ?

r^ ON ON 0) c 3

^^ ? II c e .2 ?

r?oo

O rn vOrn* ON

t/3 0> ? ?

'il C/J d?

3

O ?/S* NO

O ON ON

c o

ON* ?

8.? ?? ? ON -, o. JS ^ ? e w o C e ?tu

ON ?

NO

O

?J? ?_, C/5 ? e a> ?Dcd ? *??

o>5 S co a h e

? J= -ti *

q ^ ?D.. g * ?

p p no t-^

un m o o un oo Tf ~

Onun (N ? un oo ? o?