Coral Reefs

pp 1–14

Connecting Palau’s marine protected areas: a population genetic approach to conservation

  • Annick Cros
  • Robert J. Toonen
  • Megan J. Donahue
  • Stephen A. Karl
Report

DOI: 10.1007/s00338-017-1565-x

Cite this article as:
Cros, A., Toonen, R.J., Donahue, M.J. et al. Coral Reefs (2017). doi:10.1007/s00338-017-1565-x

Abstract

Bleaching events are becoming more frequent and are projected to become annual in Micronesia by 2040. To prepare for this threat, the Government of Palau is reviewing its marine protected area network to increase the resilience of the reefs by integrating connectivity into the network design. To support their effort, we used high-throughput sequencing of microsatellites to create genotypes of colonies of the coral Acropora hyacinthus to characterize population genetic structure and dispersal patterns that led to the recovery of Palau’s reefs from a 1998 bleaching event. We found no evidence of a founder effect or refugium where colonies may have survived to recolonize the reef. Instead, we found significant pairwise F′st values, indicating population structure and low connectivity among most of the 25 sites around Palau. We used kinship to measure genetic differences at the individual level among sites and found that differences were best explained by the degree of exposure to the ocean [F1,20 = 3.015, Pr(>F) = 0.01], but with little of the total variation explained. A permutation test of the pairwise kinship coefficients revealed that there was self-seeding within sites. Overall, the data point to the population of A. hyacinthus in Palau recovering from a handful of surviving colonies with population growth primarily from self-seeding and little exchange among sites. This finding has significant implications for the management strategies for the reefs of Palau, and we recommend increasing the number and distribution of management areas around Palau to capture the genetic architecture and increase the chances of protecting potential refuges in the future.

Keywords

Microsatellites Connectivity Oceanographic modeling Genetic architecture 

Introduction

Coral reefs are declining due to a variety of anthropogenic impacts that are exacerbated by climate change (Hughes et al. 2003; Baker et al. 2008; McLeod et al. 2009; Burke et al. 2011). As atmospheric levels of CO2 continue to rise, coral bleaching events are projected to increase in frequency and severity worldwide (van Hooidonk et al. 2013); however, the rates of increases are predicted to vary between regions (van Hooidonk et al. 2013). In the best-case greenhouse gas scenario (RPC6.0) where there is reduction and stabilization of CO2 emissions, bleaching events are predicted to occur annually by 2078 across reefs worldwide, with some areas, such as Micronesia, surpassing thermal bleaching thresholds annually as early as 2040 (van Hooidonk et al. 2013).

Maintaining connectivity among populations has become a critical goal for conservation strategies to increase the recovery of reefs affected by thermal stresses and bleaching (Jones et al. 2007, 2009; Almany et al. 2009; McLeod et al. 2009). Networks of marine protected areas (MPAs) may promote recovery from disturbance by allowing reefs to be mutually replenishing through coral recruitment (Hughes et al. 2003; Salm et al. 2006; IUCN 2008; McLeod et al. 2009). Even the best-studied and best-funded MPA networks such as the network on the Great Barrier Reef, however, did not consider connectivity between their no-take reserves when rezoned because they lacked the information (Fernandes et al. 2005). The lack of connectivity data is a challenge for most MPA networks around the world due to the high cost of producing this information for managers (Almany et al. 2009; Lagabrielle et al. 2014; Magris et al. 2014).

Connectivity of marine organisms is particularly difficult to measure due to the small size and high dispersal potential of pelagic larvae (reviewed in Levin 2006; Cowen and Sponaugle 2009). Two main approaches are generally used to determine population connectivity: (1) computer modeling to simulate the dispersal of particles forced with physical data (e.g., wind and tides) and (2) genetic data (i.e., DNA) to study gene flow between populations (Hellberg 2007; Cowen and Sponaugle 2009). Seascape genetics brings together both approaches to test for environmental drivers of spatial genetic structure (Selkoe et al. 2008, 2016; White et al. 2010; Liggins et al. 2013; Riginos and Liggins 2013), allowing cross-validation between oceanographic models and genetic data (Baums et al. 2006; Pringle and Wares 2007; White et al. 2010; Foster et al. 2012). In this study, we use genetic techniques to test whether the connectivity predicted by an oceanographic model of Palau developed by Golbuu et al. (2012) coincides with genetic differences on Palau’s reefs.

Palau is the western-most island in Micronesia. Palau’s reefs suffered high mortality during the 1998 bleaching event, and Palau has been identified as one of the areas that will be exposed to bleaching conditions annually by 2040 (van Hooidonk et al. 2013). The Government of Palau is responding to future bleaching events by reviewing the design of the network of MPAs, incorporating connectivity to increase its resilience. To inform the placement of new MPAs, Golbuu et al. (2012) used an oceanographic model built on current and wind data predicting long-distance dispersal as well as a “sticky water” model developed specifically to predict larval retention around reefs locally (Wolanski 1994; Wolanski and Spagnol 2000; Andutta et al. 2012). The long-distance model describes the potential dispersal of coral larvae from Yap to Palau. The localized model describes low self-seeding for most of the barrier reef and an export of larvae from the southern barrier reefs to the northern barrier reefs. Golbuu et al. (2012) tested their model against coral cover and density of juvenile Acropora colonies. They found that areas of high coral cover corresponded to areas of predicted high self-seeding and areas of high density of juvenile Acropora colonies corresponded to areas of predicted high density of larval recruits. Density of larval recruits was estimated by adding the projected density of larvae from self-seeding and project density of larvae coming from other sites.

To better understand how Palauan reefs recovered from the 1998 bleaching event and to compare the oceanographic dispersal model (Golbuu et al. 2012) to genetic structure, we assessed the genetic structure and genetic diversity of populations of A. hyacinthus around the reefs of Palau. Acropora hyacinthus is a table coral found on shallow barrier reefs between 3 and 10 m throughout the Indo-Pacific. Although it is one of the dominant coral species growing on Palau’s barrier reef at shallow depths (Golbuu et al. 2007; Victor et al. 2009), A. hyacinthus is rare or absent on the patch reefs, fringing reef and lagoon of Palau (Bruno et al. 2001). In 1998, A. hyacinthus suffered heavy mortality from a bleaching event in Palau, almost disappearing from the reef (Bruno et al. 2001), but by 2005 it was once more dominant on the reef slopes (Golbuu et al. 2007). Two hypotheses were put forward for this rapid recovery. The first was that Palau received an important recruitment pulse of larvae from Yap (Golbuu et al. 2012). However, Cros et al. (2016) demonstrated that there was no genetic evidence to support mass recruitment from Yap to Palau. The second hypothesis was that there were more surviving colonies than originally described by Bruno et al. (2001). Bruno et al. (2001) only surveyed a few sites on the barrier reef and may have missed areas with surviving colonies acting as a source of recruits to recolonize the reef (Golbuu et al. 2007; Victor et al. 2009; Cros et al. 2016). Our objective was to understand the patterns of dispersal that have led to recovery, to detect evidence of self-seeding and to provide recommendations to increase resilience of the reefs in Palau.

Materials and methods

Study species

Acropora hyacinthus is a hermaphroditic broadcast spawner that releases egg and sperm bundles during mass spawning events (Ayre and Hughes 2000). Although A. hyacinthus can also reproduce asexually, previous studies have found very few clones in the field (Ayre and Hughes 2000; Márquez et al. 2002). Acropora hyacinthus reaches maturity at 3–5 yr (Wallace 1985). After successful fertilization, under laboratory conditions, planktotrophic A. hyacinthus larvae (Toh et al. 2013) start settling and metamorphosing after 3 d (Toh et al. 2012) but can remain in the water column for up to 90 d before settling (Márquez et al. 2002).

Sampling locations and methodology

In February and May 2012, 25 sites on the outer barrier reef of Palau were sampled at a shallow depth (<10 m) using SCUBA (Fig. 1; Table 1). The outer barrier reef was divided into four quadrants with an east–west division along the length of the atoll and a north–south division separating the northern lagoon and surrounding reefs from the southern lagoon and the reefs adjacent to the two main islands of Babeldaob and Koror, forming four zones. Each zone is characterized by a different exposure to waves driven by the north-easterly trade winds and the Western Pacific monsoon winds. During winter (December–March), trade winds blow predominantly from the northeast with the northeast reefs most exposed to waves. In the summer, the Western Pacific monsoon winds blow from the southwest and the southwest reefs are exposed to smaller waves (Australian Bureau of Meteorology and CSIRO 2014). Within each exposure zone, sites were selected to represent a range of habitat types. Additionally, sites both within and outside MPAs were included to represent management categories found on the barrier reef. A total of 1200 1-cm3 branch tips were collected by haphazardly sampling 48 colonies of A. hyacinthus in 4 × 200 m belt transects at each of these 25 sites. Branch tips were preserved in salt-saturated DMSO at room temperature (Gaither et al. 2011).
Fig. 1

Map of Palau and its reefs with 25 sampling locations. Dashed black lines indicate the division of the reef into exposure zones. Blue lines are boundaries of marine protected areas as of 2012

Table 1

Site number, seeding category, exposure, date of collection, GPS coordinates and number of samples of Acropora hyacinthus collected at 25 sites in Palau

Site

Seeding

Exposure

Date of collection

Longitude (N)

Latitude (W)

No. samples

Self

Total

S1

Low

Med–high

SE

14.02.12

7.2874

134.50295

48

S2

Low

High

SW

18.02.12

7.5610

134.46864

47

S3

Medium

High

SW

16.02.12

7.4183

134.34557

48

S4

Medium

High

SW

17.02.12

7.3070

134.23141

47

S5

N/A

N/A

SW

20.02.12

7.0111

134.21833

44

S6

N/A

N/A

NE

11.03.12

8.0421

134.68630

46

S7

Low

Med–high

SW

22.02.12

7.2523

134.22093

45

S8

Low

Med–high

SE

21.05.12

7.2619

134.54426

48

S9

Low

Med–high

SE

21.05.12

7.3623

134.61971

48

S10

Low

High

SE

22.05.12

7.1113

134.36692

48

S11

Low

Med–low

NE

23.05.12

7.9902

134.65965

48

S12

Low

Low

NE

23.05.12

7.9886

134.70319

48

S13

Low

Low

NE

24.05.12

7.8788

134.68135

48

S14

Low

Low

NE

25.05.12

7.8150

134.66043

47

S15

High

Med–High

NE

26.05.10

7.6678

134.64970

48

S16

Low

Med–high

NE

26.05.11

7.5860

134.64929

48

S17

Low

Med–high

SE

26.05.12

7.4297

134.64202

48

S18

Low

Med–high

SW

28.05.12

7.0796

134.26157

48

S19

Low

High

NW

29.05.12

7.7224

134.56752

48

S20

Low

Med–high

NW

31.05.12

8.0014

134.53610

48

S21

Low

Med–high

SE

01.06.12

7.0556

134.31810

48

S22

Low

Med–high

NW

02.06.12

7.8602

134.50802

48

S23

Low

Med–low

SE

04.06.12

7.1633

134.41277

48

S24

Low

High

SW

05.06.12

7.5307

134.40110

48

S25

Low

Med–low

NW

06.06.12

7.8018

134.50800

48

Self-seeding categories are defined based on levels of self-seeding in the Golbuu et al. (2012) model as follows: high (60–65%); medium (16–25%); low (0–15%). Total seeding groups are: high (61–90%); medium–high (41–60%); medium–low (21–40%); low (0–20%)

DNA extraction and sequencing

A detailed description of DNA extraction and sequencing is described in Cros et al. (2016). Briefly, genomic DNA was isolated from each branch tip, and 18 microsatellite loci (Electronic Supplementary Material, ESM, Table S1) were amplified using a forward primer with a short tag to create 48 unique colony IDs for each microsatellite locus (ESM Table S2). The barcoded amplicons were pooled by site. An Illumina adaptor (Illumina Inc., Hayward, CA, USA) was ligated to generate a library with the following unique structure: siteID-colonyID-forwardprimer-flankingregion-tandemrepeats-flankingregion-reverseprimer. Each library was sequenced on an Illumina MiSeq.

Data processing

Raw sequences were processed following the bioinformatics pipeline in Cros et al. (2016). In brief, the sequences were demultiplexed by site, merged, separated according to primer and colony ID, and trimmed for low-quality sequences. They were then collapsed into unique sequences and counted. To eliminate PCR and sequencing artifacts, a set of filters was developed in Python (https://github.com/annickcros/Ahyacinthus-filters.git). Flanking regions were filtered from simple tandem repeats (STR) using emboss: etandem (Rice et al. 2000). Genotypes were created based on the STR. Data were transformed to genodive v. 2.0b27 (Meirmans and van Tienderen 2004) file format using formatting as described in Cros et al. (2016). After removing loci with over 15% missing data, we used 11 loci (Table 2) to genotype colonies. The final number of colonies analyzed for each locus varies between 37 and 48 per site.
Table 2

Number of alleles (A), observed heterozygosity (HO) with standard deviation (SD) and range of length (nt) of microsatellite loci

Locus

A

HO (SD)

nt

Locus 1

2

0.34 ± 0.10

28–36

Locus 3

6

0.69 ± 0.07

21–36

Locus 4

20

0.91 ± 0.02

15–75

Locus 5

11

0.55 ± 0.05

57–90

Locus 6

4

0.50 ± 0.11

36–45

Locus 8

14

0.83 ± 0.03

39–108

Locus 11

8

0.66 ± 0.06

12–44

Locus 12

17

0.83 ± 0.05

36–93

Locus 13

8

0.72 ± 0.08

54–81

Locus 14

12

0.68 ± 0.07

20–84

Locus 16

7

0.32 ± 0.24

32–56

Analyses

Population differentiation

We first tested for clones using genodive v. 2.0b27. To characterize the genetic structure of each of the 25 sites, we assumed that each site was a single population and calculated in genodive the number of alleles, the effective number of alleles, and indices of genetic diversity, as well as observed (HO) and expected (HE) heterozygosities and inbreeding coefficient GIS (Table 3) at each site. To test for overall population structure among the 25 sampled sites, we calculated global FST, F′ST and FIS and corresponding p values with an AMOVA in genodive (Table 4). Indices of global genetic diversity, including observed (HO), expected (HE) and corrected heterozygosities (H′T), inbreeding coefficient (GIS) and Nei’s fixation index GST, were also calculated (Table 4).
Table 3

Indices of genetic diversity of Acropora hyacinthus at each of 25 sites around Palau, including number of alleles (N), effective number of alleles (NE), expected (HE) and observed (HO) heterozygosity, inbreeding coefficient (GIS) and significance levels of inbreeding coefficients (p)

Site

N

NE

HO

HE

GIS

p value

S1

8.36

4.35

0.27

0.66

0.60

<0.01

S2

7.27

3.99

0.31

0.65

0.53

<0.01

S3

7.46

4.03

0.31

0.66

0.52

<0.01

S4

7.64

4.15

0.35

0.66

0.48

<0.01

S5

6.36

3.71

0.27

0.66

0.59

<0.01

S6

7.64

4.00

0.28

0.66

0.57

<0.01

S7

6.64

3.70

0.22

0.65

0.67

<0.01

S8

7.55

4.35

0.37

0.65

0.43

<0.01

S9

7.64

3.92

0.40

0.63

0.37

<0.01

S10

6.64

3.56

0.37

0.62

0.42

<0.01

S11

8.55

3.87

0.40

0.61

0.35

<0.01

S12

7.55

3.39

0.45

0.58

0.22

<0.01

S13

7.46

3.55

0.49

0.60

0.19

<0.01

S14

8.09

4.42

0.37

0.66

0.44

<0.01

S15

7.46

3.98

0.33

0.64

0.48

<0.01

S16

8.09

3.74

0.41

0.61

0.32

<0.01

S17

6.46

2.94

0.49

0.52

0.06

<0.01

S18

7.09

3.56

0.48

0.57

0.17

<0.01

S19

6.90

3.54

0.37

0.58

0.36

<0.01

S20

6.91

3.81

0.48

0.58

0.18

<0.01

S21

7.55

3.86

0.44

0.59

0.25

<0.01

S22

5.73

3.20

0.52

0.52

0.01

0.22

S23

5.46

2.91

0.51

0.51

−0.01

0.55

S24

7.73

3.72

0.23

0.63

0.63

<0.01

S25

6.55

3.39

0.53

0.55

0.04

<0.01

Table 4

Measures of population diversity and differentiation of Acropora hyacinthus at 25 sites around Palau calculated from 1188 individual colonies and 11 loci

Index

Value

p value

FST

0.30

<0.01

F′ST

0.08

<0.01

FIS

0.30

<0.01

N

13.09

 

NE

3.54

 

HO

0.90

 

HE

0.10

 

H′T

0.30

 

GIS

0.37

 

G′ST

0.03

<0.01

Standard (FST) and corrected (F′ST) fixation indices, inbreeding coefficient (FIS), number of alleles (N), effective number of alleles (NE), observed (HO) and expected (HE) heterozygosity, corrected heterozygosity (H′T), Nei’s inbreeding (GIS) and corrected (G′ST) fixation coefficients and significant levels when appropriate

To test the hypothesis that A. hyacinthus recovered from a few individuals surviving the 1998 bleaching event, we looked for evidence of recent bottlenecks or founder effects using bottleneck 1.2.02 (Piry et al. 1999). We used the graphical test from Luikart et al. (1998) based on a mode shift away from an L-shaped distribution of allelic frequencies. We used the Wilcoxon signed-rank test (10,000 iterations) using both a two-phase mutational model (TPM, incorporating 70% stepwise and 30% multistep mutations) and an infinite allele (IAM) mutational model due to the relatively small number of microsatellite loci scored in our dataset.

To test for connectivity between sites and patterns of genetic structure, we used pairwise differentiation tests between sites in genodive and reported F′ST and p values in Table 5. Jost’s D values were also estimated (ESM Table S3). We also looked for patterns of genetic structure by performing a principal component analysis (PCA) in genodive (Meirmans and van Tienderen 2004) on sites using a covariance matrix of allele frequencies with 10,000 permutations. The graphs were plotted in R v. 3.2.3 (R Core Team 2015; Fig. 2).
Table 5

Pairwise F′ST comparisons for 25 sites in geographic order around the barrier reef of Palau (top diagonal) and permutation p values (lower diagonal). Values in italics are not statistically significant

 

S21

S10

S23

S1

S8

S9

S17

S16

S15

S14

S13

S12

S11

S6

S20

S22

S25

S19

S2

S24

S3

S4

S7

S18

S5

S21

0.04

0.05

0.05

0.04

0.02

0.04

0.03

0.04

0.04

0.05

0.06

0.01

0.06

0.00

0.04

0.03

−0.02

0.12

0.04

0.12

0.12

0.07

0.02

0.11

S10

0.00

0.08

0.05

0.03

0.05

0.11

0.06

0.05

0.05

0.07

0.07

0.05

0.08

0.04

0.05

0.06

0.02

0.13

0.06

0.11

0.10

0.09

0.08

0.09

S23

0.00

0.00

0.15

0.14

0.10

0.06

0.12

0.14

0.11

0.11

0.11

0.09

0.14

0.04

0.03

0.00

0.07

0.20

0.12

0.24

0.23

0.19

0.10

0.19

S1

0.00

0.00

0.00

0.01

0.05

0.15

0.06

0.03

0.02

0.05

0.10

0.05

0.04

0.07

0.13

0.11

0.07

0.08

0.02

0.02

0.04

0.02

0.11

0.04

S8

0.00

0.01

0.00

0.16

0.05

0.15

0.04

0.01

0.02

0.03

0.10

0.05

0.04

0.05

0.11

0.10

0.03

0.11

0.00

0.07

0.07

0.03

0.09

0.08

S9

0.05

0.00

0.00

0.00

0.00

0.07

0.04

0.05

0.06

0.09

0.09

−0.02

0.08

0.02

0.07

0.07

0.01

0.15

0.07

0.10

0.08

0.07

0.02

0.09

S17

0.00

0.00

0.00

0.00

0.00

0.00

0.09

0.13

0.14

0.16

0.09

0.07

0.16

0.06

0.08

0.05

0.05

0.22

0.15

0.22

0.21

0.18

0.04

0.19

S16

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.02

0.07

0.07

0.11

0.04

0.09

0.05

0.11

0.09

0.06

0.14

0.06

0.08

0.08

0.06

0.05

0.11

S15

0.00

0.00

0.00

0.03

0.16

0.00

0.00

0.05

0.06

0.03

0.08

0.05

0.07

0.07

0.11

0.12

0.04

0.09

0.04

0.07

0.09

0.05

0.08

0.12

S14

0.00

0.00

0.00

0.07

0.06

0.00

0.00

0.00

0.00

0.04

0.12

0.05

0.02

0.05

0.12

0.07

0.05

0.08

−0.01

0.09

0.06

0.05

0.11

0.05

S13

0.00

0.00

0.00

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.11

0.09

0.08

0.06

0.10

0.07

0.05

0.11

0.02

0.12

0.11

0.08

0.11

0.14

S12

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.07

0.11

0.08

0.11

0.10

0.05

0.19

0.12

0.16

0.18

0.14

0.11

0.14

S11

0.06

0.00

0.00

0.00

0.00

0.98

0.00

0.00

0.00

0.00

0.00

0.00

0.08

0.04

0.08

0.07

0.02

0.15

0.07

0.11

0.10

0.09

0.04

0.11

S6

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.00

0.08

0.00

0.00

0.00

0.09

0.15

0.12

0.07

0.09

0.03

0.09

0.06

0.05

0.14

0.05

S20

0.54

0.00

0.00

0.00

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.04

0.02

0.00

0.16

0.04

0.16

0.13

0.11

0.03

0.12

S22

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.02

0.04

0.22

0.11

0.20

0.19

0.15

0.05

0.19

S25

0.00

0.00

0.59

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.02

0.00

0.04

0.17

0.08

0.20

0.18

0.15

0.06

0.15

S19

0.99

0.09

0.00

0.00

0.02

0.21

0.00

0.00

0.01

0.00

0.00

0.00

0.08

0.00

0.53

0.00

0.00

0.16

0.04

0.15

0.13

0.09

0.01

0.13

S2

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.11

0.09

0.12

0.09

0.20

0.11

S24

0.00

0.00

0.00

0.04

0.57

0.00

0.00

0.00

0.01

0.75

0.05

0.00

0.00

0.03

0.00

0.00

0.00

0.03

0.00

0.10

0.08

0.05

0.11

0.09

S3

0.00

0.00

0.00

0.07

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.03

0.17

0.06

S4

0.00

0.00

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.18

0.02

0.14

0.03

S7

0.00

0.00

0.00

0.09

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.01

0.06

0.15

0.12

0.02

S18

0.03

0.00

0.00

0.00

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.12

0.00

0.00

0.00

0.00

0.00

0.18

S5

0.00

0.00

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.03

0.14

0.00

Nonsignificant F′ST values are italics

Fig. 2

Principal component analysis of Acropora hyacinthus colonies at 25 sites around Palau using a covariance matrix of allele frequencies with 10,000 permutations. The first axis explains 31.9% of the variation and the second axis explains 14.9% of the variation among sites. Sites are grouped by exposure zones (Table 7)

To examine spatial genetic structure of A. hyacinthus, we carried out a Bayesian clustering algorithm implemented in structure ver. 2.3.4 (Pritchard et al. 2000). We used a no-admixture model with location as a prior and a burn-in of 10,000 chains followed by 10,000 MCMC replications as suggested by Benestan et al. (2016). Twenty independent runs were carried out for each number of clusters (K) from 1 to 25. The most likely value of K was evaluated using the method of Evanno et al. (2005) in clumpak (Kopelman et al. 2015).

Patterns of connectivity

We created a pairwise matrix of geographic distances by measuring the shortest distance between each pair of sites following the contour of the barrier reef using esri arcgis v.10.2.2. (ESM Table S4). We tested for isolation-by-distance with a Mantel test (Mantel 1967) in genodive comparing a matrix of transformed pairwise F′ST values [F′ST/(F′ST – 1)] with a matrix of log-transformed geographic distances with 20,000 permutations.

To compare the dispersal patterns predicted by Golbuu et al. (2012) with genetic differentiation, we overlaid the sampling sites from this study with the map of self-seeding rates (the percentage of particles remaining in the site after 120 h) and the map of total seeding rate (the sum of particles retained at a given site plus the imports from other release sites after 120 h) from Golbuu et al. (2012). Sites were grouped according to the level of self-seeding and total seeding on the barrier reef. For self-seeding, we used the following groups (Table 1): high (60–65%), 1 site; medium (16–25%), 3 sites; and low (0–15%), 19 sites. For total seeding, we used the following groups: high (61–90%), 6 sites; medium–high (41–60%), 11 sites; medium–low (21–40%), 3 sites; and low (0–20%), 4 sites. We performed a hierarchical AMOVA in genodive using these groupings. We also performed a group comparison test in genodive that tests whether groups of populations differed in their values of certain summary statistics. The group comparison test calculates summary statistics for each group and then uses a permutation test to test for differences between the groups. The OSx statistic was used (Goudet 1995), which is the sum of the squared differences in the test statistic over all pairwise combinations of groups. Permutations take place by randomizing the populations over the groups (Meirmans and van Tienderen 2004). We compared HO and HE and indices of genetic diversity GIS (inbreeding coefficient) and GST (fixation index) to compare the self-seeding and total seeding groups (Table 6).
Table 6

Average of indices of genetic diversity of Acropora hyacinthus within sites grouped by categories of self-seeding and total seeding, and permutation test for differences among the groups

Coefficient

Exposure

OSx

p value

Low

Medium

High

Self-seeding

 Ho

0.39

0.38

0.33

0.07

0.81

 He

0.61

0.63

0.64

0.04

0.82

 Gis

0.36

0.40

0.48

0.15

0.84

 Gst

0.03

0.03

N/A

0.04

0.59

 

Low

Med/low

Med/high

High

  

Total seeding

 Ho

0.43

0.48

0.39

0.32

0.29

0.10

 He

0.61

0.56

0.60

0.64

0.13

0.21

 Gis

0.29

0.15

0.35

0.49

0.62

0.10

 Gst

0.03

0.02

0.02

0.03

0.02

0.91

OSx-statistic is used to test for significance (Goudet 1995)

We repeated the methods described above, performing a hierarchical AMOVA and a group comparison in genodive to test for a genetic division between the sites grouped by western and eastern reefs, by southern and northern reefs and by exposure zones characterized by the degree of exposure to wind and waves: northeast, northwest, southeast and southwest reefs (Fig. 1; Tables 1, 7).
Table 7

Average of indices of genetic diversity of Acropora hyacinthus within sites grouped by exposure zones (SE southeast, SW southwest, NE northeast, NW northwest), and permutation test for differences among the groups

Coefficient

Exposure zones

OSx

p value

SE

SW

NE

NW

Ho

0.41

0.31

0.39

0.48

0.24

0.02

He

0.60

0.64

0.62

0.56

0.13

0.02

Gis

0.32

0.52

0.37

0.14

0.54

0.01

Gst

0.03

0.03

0.02

0.01

0.03

0.27

OSx-statistic is used to test significance (Goudet 1995)

We tested for self-seeding by calculating pairwise kinship coefficients of individuals within sites and between sites using genodive and compared distributions of within-site and between-site pairwise coefficients. To test for differences in mean pairwise kinship between within-site and between-site pairs (Fig. 3a), we resampled without replacement (resample size = 27,646, number of resamples = 1000) from the distribution of all pairs and calculated the mean difference between within-site pairs and between-site pairs (Fig. 3b). Sites with high within-site pairwise kinship coefficients indicate sites with high self-seeding, and sites with high between-site pairwise kinship coefficients indicate sites with high total seeding. We ranked the mean pairwise kinship coefficients within and among sites and compared them with the predicted self-seeding and total seeding groups as defined by Golbuu et al. (2012) (ESM Table S5).
Fig. 3

a Histograms of the distribution of the frequency of pairwise kinship coefficients. In blue, distribution of 27,646 pairwise kinship coefficients within sites. In red, distribution of 677,432 pairwise kinship coefficients between sites. Histograms are plotted with a normalized axis. b Distribution of the mean differences of pairwise kinship coefficients within and between sites generated by permutation (1000)

We used a canonical analysis of principal coordinates (CAP) to see which of the following predictor variables explained the most genetic variation: distance (ESM Table S4), differences between northwest exposure, east–west exposure and their interaction. We used ANOVA to test for the importance of the predictors. All analyses were performed in package vegan 2.3-5 in R v. 3.2.3 (R Core Team 2015).

Results

Population differentiation

We genotyped between 44 and 48 colonies at each site (Table 1) for each of our 11 microsatellite loci. No clones were detected; therefore, downstream analyses were performed on all samples. At each site, the effective number of alleles per locus varied between 3 and 4.5 (Table 3). Observed heterozygosity (HO) ranged from 0.22 to 0.53 and expected heterozygosity (HE) from 0.51 to 0.66 and all but two were out of Hardy–Weinberg equilibrium (HWE; Table 3).

Observed heterozygosity averaged over all sites and loci (HO = 0.39 ± 0.06, SD) showed a lower value than the expected heterozygosity averaged over all sites and loci (HE = 0.61 ± 0.079) and a higher inbreeding coefficient than expected (GIS = 0.37, p < 0.01). Both the fixation index FST and the standardized GST (Nei) were low but significant (p < 0.01), indicating fine-scale population structure among the 25 sites around Palau (Table 4). Likewise, pairwise F′ST comparisons indicated that most of the 25 sites around Palau were significantly different from each other (Table 5). Pairs of sites that were not significantly different were not organized in any obvious geographic patterns.

The base-shift graphical test for bottleneck showed normal L-shaped distributions for all 25 sites, indicating no evidence of a bottleneck for A. hyacinthus around Palau.

The results from structure and clumpak indicated the most likely number of clusters was four. Summary graphs, however, did not show a clear geographic separation (ESM Fig. S1). Similarly, the PCA did not highlight any obvious spatial patterns or geographic division among sites, with all sites distributed fairly evenly along the first and second axis (Fig. 2).

Patterns of connectivity

We found no evidence of isolation-by-distance. There was no significant relationship between either the untransformed (Spearman’s r = 0.024, p = 0.270) or log–log-transformed (Spearman’s r = −0.0038, p = 0.469) geographic and genetic distances.

We overlaid the sampling sites from this study with the map of self-seeding rates and the map of total seeding rate from Golbuu et al. (2012). We tested for a significant difference between sites grouped according to the self-seeding and total seeding classes defined in Golbuu et al. (2012). The hierarchical AMOVA in genodive did not indicate significant genetic differences (F′st) between sites grouped by self-seeding classes or the total seeding classes, and we found FCT (SELF) = 0.001 (p = 0.37) and FCT (total) = 0.003 (p = 0.12). Similarly, the OSx-statistic of the group comparison test in genodive did not reveal any significant differences in either the value of observed or expected heterozygosities, inbreeding coefficient or fixation index between sites with high, medium or low self-seeding rates or between sites with low, medium–low, medium–high or high total seeding rates (Table 6).

The AMOVA hierarchical analysis for sites grouped into north, south, east and west did not show any significant difference between groups. However, there were significant differences among sites grouped by exposure zones (FCT = 0.006, p < 0.05). In addition, the group comparison for the exposure zones showed significant differences in observed and expected heterozygosities and inbreeding coefficients (Table 7) with the sites grouped in the northwest exposure zone showing the lowest inbreeding coefficients and driving the observed difference.

The overlaid histograms of the frequency distribution of pairwise kinship coefficients within and between sites indicated that there were higher kinship coefficients within sites than between sites (Fig. 3a). We plotted the results of the permutation analysis, testing the significance of the difference between the mean of pairwise kinship coefficients within and between sites (Fig. 3b), and noted that the distribution of the mean differences between permuted within- and between-site pairs were squarely distributed around zero. The value of the mean pairwise kinship coefficient within sites was 0.0286 (95% CI 0.0013 to 0.0011), between sites was −0.0012 (95% CI −0.0001 to −0.00002), and the difference in means was 0.0298 (95% CI −0.0012 to 0.0012). The within- and between-site distribution fell outside of the 95% percentile indicating that the mean of pairwise kinship coefficients for within and between sites was significantly different.

The four sites with the highest within-site mean kinship coefficients were S17 and S23 in the southeast and S2 and S3 in the southwest (ESM Table S5). This does not correspond to the class given by Golbuu et al. (2012) where S17, S23 and S2 were classed as “low” for self-seeding. The four pairs of sites with highest between-site mean kinship coefficients were S23 and S25, S3 and S4, S22 and S23, and S4 and S5 (ESM Table S5). This closely approximates predictions by Golbuu et al. (2012) where S4, S5 are classed as “high” for total seeding and S22 as “medium–high.” Overall kinship coefficients, self-seeding and total seeding classes do not match.

Variation in genetic differentiation between sites was best explained by exposure zones [ANOVA, F1,20 = 3.015, Pr(>F) = 0.01] and explained 31% of the of the total variation.

Discussion

Understanding the processes by which A. hyacinthus has recovered from the devastation of the 1998 bleaching event is critical for the future management of these reefs, for better placement of new MPAs and for protecting areas that may act as refugia. The hypothesis that recovery came from a massive influx of recruits from the island of Yap (Golbuu et al. 2012) is inconsistent with every site being as differentiated from each other as they are from Yap (Cros et al. 2016). Here, we tested the hypothesis that recovery was generated from a few surviving colonies from Palau and further compared the modeled larval dispersal from Golbuu et al. (2012) to determine how well it predicted the observed genetic diversity among sites in Palau. We present a series of hypotheses on the processes that resulted in the genetic patterns we observe and propose different explanations for the recovery of the reef.

Founder effect

If there had been a recolonization of the population of A. hyacinthus on the barrier reef from only a few colonies that had survived the 1998 bleaching event, there should be signs of founder effects in the new populations. Yet the bottleneck analysis shows no disproportionate distribution of allelic frequencies. One possible explanation is that there were more surviving colonies than we expected eliminating any evidence of a founder effect. A review by Dlugosch and Parker (2008) reported that among 11 cases of intentional introduction, loss of genetic diversity was detected in all but one case involving the introduction of fewer than 250 individuals. In each case, representing a variety of taxa (e.g. birds, reptiles and insects), the number of individuals was known confidently and derived from a single source population. In a study of the introduction of groupers in Hawaii, Gaither et al. (2010) found signs of a bottleneck for introductions of fewer than 750 adults. Genetic diversity in Palau is equivalent to other sites that did not suffer dramatic coral loss (Cros et al. 2016). Given results in other species, this suggests that hundreds of individuals likely survived.

Self-seeding and differential survival

In addition to the lack of a bottleneck effect, we found that most of the 25 sites around the barrier reef of Palau display genetic differences with significant F′ST for the majority of the pairwise comparisons (Table 5). At the same time, the distribution of kinship coefficients within and between sites shows evidence of self-seeding (Fig. 3). A possible hypothesis to explain the processes that drive the genetic diversity of A. hyacinthus around Palau’s barrier reef is that enough colonies survived the 1998 bleaching to recolonize the reef and that their larvae mostly recruited close to parental colonies creating genetic structure between sites. Another possible explanation is that there was differential survival of recruits such that even with some gene flow, individuals at a site grew to be differentiated through time (Toonen and Grosberg 2011; Gorospe and Karl 2015). An alternative explanation is that we sampled cryptic species (Ladner and Palumbi 2012), but there was no evidence for linkage groups in either the structure analysis (ESM Fig. S1) or the PCA (Fig. 2).

Reef patchiness and clumped recruitment

Pinsky et al. (2012) modeled the effect of habitat patchiness on population openness and tested it on different reefs. They found that in many cases, habitats were sufficiently patchy at scales of tens of kilometers to create largely closed populations, especially for species with low dispersal abilities. Habitat patchiness can be the result of disturbance, in particular coral mortality due to bleaching (Andréfouët et al. 2002; Hughes et al. 2003). In the case of Palau, it is likely that colonies of A. hyacinthus survived in patches. Although A. hyacinthus larvae have the potential for long-distance dispersal, there is increasing evidence that, like other marine organisms, the realized dispersal is considerably more restricted (Hughes et al. 2000; Levin 2006; Shanks 2009; D’Aloia et al. 2013). Under this scenario, the expectation of partially closed populations on the reef of Palau matches the results of F′ST and kinship coefficients. The existing patches mostly grow through self-seeding with a few larvae being exported and starting new patches that are not completely closed and therefore mixing over several generations. Alternatively, there are larval exchanges over longer distances, but sibling larvae are all transported in a batch and settle together, regardless of where they come from, creating genetic patchiness (Selkoe et al. 2006; Bernardi et al. 2012; Iacchei et al. 2013). These chaotic patterns of spatial genetic structure are common in marine species; a number of hypotheses have been proposed (Eldon et al. 2016), but would require further study to evaluate the underlying mechanisms.

Oceanographic influence

We tested the genetic patterns we observed against the predictions of the oceanographic larval dispersal model developed by Golbuu et al. (2012). The model predicted that the southern reefs contribute most to total seeding, in particular of the northern reef, which exports most of its larvae early on to open ocean. In contrast to the predictions of their model, we saw no evidence of increased relatedness in areas defined as high self-seeding (Table 6). We saw an increase in heterozygosity and decrease in inbreeding coefficients, however, for sites grouped in the northwest exposure zone (Table 7), which would indicate that the northern reefs receive inputs from other sites as opposed to the southern reefs that show evidence of self-seeding. Regardless of the process, physical drivers appear responsible because the four different exposure zones best-predicted the genetic difference between distances, north–south and east–west divisions.

Management implications

Our study supports previous observations by Victor et al. (2009) and Golbuu et al. (2007) who argued that recovery was most likely due to some surviving colonies rather than the catastrophic mortality reported by Bruno et al. (2001). There are two likely scenarios why these colonies survived. The first is that the colonies were in an area that provided refugia to thermal stressors, such as a locally cooling current or on a slope that provided shading. In this case, the surviving colonies would not be more resistant than the colonies that died and the growth of the reef from self-seeding would not result in a more resilient reef. The second scenario is that the colonies that survived were more resistant to thermal stresses or were previously exposed to non-lethal thermal stress anomalies leading to higher adaptive capacity, in which case self-seeding would result in a more resilient reef (Thompson and van Woesik 2009; Mumby et al. 2011; Oliver and Palumbi 2011; Bahr et al. 2015).

In the first case scenario of refugia, conservation agencies recommend that resilient MPA networks should be designed to capture and protect as many of these refugia as possible (Salm et al. 2006; McLeod et al. 2009; Chollett and Mumby 2013). In the case of Palau, this recommendation would result in increasing the number of MPAs along the barrier reef so that reefs in each of the wave-exposure gradients are protected. In the second case scenario, where surviving colonies are resistant to thermal stresses, conservation efforts should focus on protecting these populations and their adjacent habitat from other stressors such pollution or overfishing and enable recruitment as much as possible (Selig et al. 2012). If only a few surviving colonies are broadly distributed, however, they may well fall outside of MPAs, and incorporating these colonies into a management plan to protect them and their surrounding habitat would require increasing the number and size of MPAs. Our data support an alternative strategy to increase the likelihood of survival in the case of repeated thermal stress events. To allow for recovery from these rare surviving colonies, it is important to have the best habitat possible for recruitment and growth. One solution to support good habitat outside of MPAs is to manage the reef outside of the existing no-take areas and reserves to maintain key ecological functions of the reef and allow for recruitment (Steneck et al. 2009).

To support coral recruitment, the key is to maintain a substrate conducive to larvae settling and surviving. Some essential elements to maintain recruitment include controlling for algae, sedimentation and water quality (Kuffner et al. 2006; Mumby et al. 2007; Burke et al. 2011). Palau already has restrictions on fishing, and the coral reef ecosystem is functional and resilient (Victor et al. 2009). Thus, although the protection of herbivores can be a good strategy in some areas (Lewis 1986; Hughes et al. 2007; McClanahan et al. 2011), it may not be the case for Palau, where there is no record of coral recruitment being hampered by algae with the exception of an algal bloom after a typhoon in 2013 which resulted in a temporary coral recruitment failure at a local level (Doropoulos et al. 2014). In contrast, the negative impacts of sedimentation and terrestrial runoff on coral reefs are well known (Hughes 1994; Bellwood et al. 2004) and have been documented as major anthropogenic impacts in Palau (Golbuu et al. 2007, 2011; Golbuu 2011). Controlling these land-based stressors may be a good strategy for maintaining the resilience of the reefs of Palau.

If the predictions of Golbuu et al. (2012) were accurate, we would expect some sites on Palau to match the genetic signature of Yap. Instead, we found that every site on Palau was equally differentiated from Yap (Cros et al. 2016) indicating that if larvae came from Yap, there were too few to leave a genetic signature. Investing in a region-wide marine protected area network to connect Yap to Palau would not enhance the resilience of the reefs of Palau to the extent desired by the Palauan government. This study highlights that in the case of Palau, management strategies that protect resistant colonies regardless of their location and habitats for successful recruitment will have the greatest impact.

Acknowledgements

We are grateful to M Belcaid and Y Cros for assistance with raw data processing and to the following University of Hawai‘i undergraduates for help in the laboratory: I Buffenstein, G Ciszek, B Haun, K Kaneshiro, M Keliipuleole, H Lim, K Niimoto, A Sifrit and T Whitman. We also thank M Iacchei for help with the manuscript and K Edwards for help with the analysis. Special thanks to The Nature Conservancy and Palau International Coral Reef Center for enabling the fieldwork and shipping permits. All collections were done under CITES permit PW 12-091 and a Palau Marine Research Permit RE-12-27. Funding was provided to A Cros and SA Karl by the Disney Wildlife Conservation Fund, to A Cros by the Graduate Women in Science Adel Lewis Grant Fellowship, the Founder Region Fellowship, the Ecology Evolution Conservation Biology Watson T. Yoshimoto grant and the Colonel Willys E. Lord Scholarship Award and the National Science Foundation grant OCE 12-60169 to RJ Toonen. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. We also thank the HIMB NSF-EPSCoR Core Genetics Lab Facility (NSF, EPS-0903833). This is the Hawai‘i Institute of Marine Biology contribution # 1677 and the School for Ocean and Earth Science and Technology contribution # 9906.

Supplementary material

338_2017_1565_MOESM1_ESM.pdf (21.7 mb)
Fig. S1 clumpak summary plots of Bayesian clustering algorithm of the genotypes of colonies at 25 sites in Palau implemented in structure with a no-admixture model with location as a prior for K = 2 to 4 (PDF 22268 kb)
338_2017_1565_MOESM2_ESM.docx (53 kb)
Supplementary material 2 (DOCX 53 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2017

Authors and Affiliations

  • Annick Cros
    • 1
  • Robert J. Toonen
    • 1
  • Megan J. Donahue
    • 1
  • Stephen A. Karl
    • 1
  1. 1.Hawai‘i Institute of Marine BiologyUniversity of Hawai‘i, MānoaKāne‘oheUSA

Personalised recommendations