Abstract
Disease, ocean warming, and pollution have caused catastrophic declines in the cover of living coral on reefs across the Caribbean. Subsequently, reef-building corals have been replaced by invertebrates and macroalgae, leading to changes in ecological functioning. We describe changes in benthic community composition and cover at 15 sites across the Belizean Barrier Reef (BBR) following numerous major disturbances—bleaching, storms, and disease outbreaks—over the 20-year period 1997–2016. We tested the role of potential drivers of change on coral reefs, including local human impacts and ocean temperature. From 1997 to 2016, mean coral cover significantly declined from 26.3% to 10.7%, while macroalgal cover significantly increased from 12.9% to 39.7%. We documented a significant decline over time of the reef-building corals Orbicella spp. and described a major shift in benthic composition between early sampling years (1997–2005) and later years (2009–2016). The covers of hard-coral taxa, including Acropora spp., M. cavernosa, Orbicella spp., and Porites spp., were negatively related to marine heatwave frequency. Only gorgonian cover was related, negatively, to our metric of the magnitude of local impacts (the Human Influence Index). Changes in benthic composition and cover were not associated with local protection or fishing. This result is concordant with studies throughout the Caribbean that have documented living coral decline and shifts in reef-community composition following disturbances, regardless of local fisheries restrictions. Our results suggest that benthic communities along the BBR have experienced disturbances that are beyond the capacity of the current management structure to mitigate. We recommend that managers devote greater resources and capacity to enforce and expand existing marine protected areas and that government, industry, and the public act to reduce global carbon emissions.
INTRODUCTION
Coral reefs worldwide have experienced remarkable changes over the past 50 years, particularly the widespread declines of reef-building corals and large, predatory fishes (1–7). These changes have caused a reduction in or effective loss of essential ecological functions, including the provisioning of habitat for fisheries production and the maintenance of reef structure for shoreline protection (8,9). Given the substantial economic and cultural value of healthy reefs (10), this degradation is affecting coastal human communities that depend on reefs for food, income, and protection from storms.
Numerous factors are responsible for the well-documented degradation of Caribbean reefs. Acroporid corals, which dominated Caribbean reefs for millions of years, experienced 90–95% mortality due to white-band disease in the 1980s (11). This disease, likely exacerbated by ocean warming (12), coupled with increased frequency and intensity of hurricanes (13–15), reduced the habitat complexity, or rugosity, of Caribbean reefs (16). Several other disease syndromes have greatly reduced the cover of other coral taxa, including black-band disease, which primarily affects brain corals (17), yellow-band disease, which primarily affects Orbicella spp. (18), and, more recently, stony coral tissue loss disease, which affects numerous species, including Dendrogyra cylindrus, Pseudodiploria strigosa, Meandrina meandrites, Eusmilia fastigiata, Siderastrea siderea and Diploria labyrinthiformis (19). Coral bleaching and other manifestations of ocean warming, including increased disease severity, are primary causes of coral loss in the Caribbean (20–27). On local scales, increased sedimentation from coastal development affects coral reefs by increasing turbidity and smothering corals (28,29). Secondary drivers include factors that have increased the cover of fleshy macroalgae (seaweeds), including the death of scleractinian corals and the consequent opening of space and other resources (30), nutrient loading, and the loss of herbivores, particularly the sea urchin Diadema antillarum due to a regional disease outbreak (31), and herbivorous fishes due to fishing (32–37).
Despite the clear and well-documented changes to Caribbean reefs, there is ongoing disagreement about the causes of and best remedies for reef decline (20,38–41). The crux of the debate is about the relative importance of local causes—pollution, eutrophication, fishing, and consequent seaweed blooms—compared with regional-to-global causes such as ocean warming and acidification. Scientists, agencies, and organizations that view localized drivers as predominant generally argue for local mitigation, the primary recommendation being fisheries restrictions, such as within Marine Protected Areas (MPAs) (34,42–44). In contrast, the view that anthropogenic climate change has been a significant or predominant cause of reef decline leads to the conclusion that without rapid cuts in carbon emissions, local protections and other localized management actions, such as restoration, will ultimately fail (20,39,45).
The purpose of this study was to measure changes to benthic communities of the Belizean Barrier Reef (BBR) from 1997 to 2016 and to determine whether they were related to protection status, fishing, local human impacts, and ocean-temperatures anomalies (i.e., ocean heatwaves). We performed surveys of the coral reef benthos at 15 sites between 1997 and 2016 (46–48). We found that benthic-community composition changed substantially during this period, and that the observed loss of corals was negatively related to ocean heatwaves and largely unaffected by local impacts, fishing or protection status.
MATERIALS AND METHODS
Study area
Scientists have tracked reef community composition across Belize for over 50 years, mostly in short-term, longitudinal studies (e.g., 11,46,48–50). Belize has an extensive, 30-plus-year-old MPA network (46) and a history of frequent large-scale disturbances (Table 1). We surveyed fore-reef benthic communities at 15–18 m depth at 15 sites along the BBR during the summer months in 1997, 1999, 2005, 2009, and 2016 (Figure 1; Table S1). Due to logistical and resource constraints, only three of the 15 sites were surveyed every year: Bacalar Chico, Middle Caye, and Tacklebox (Table S1). Study sites were selected to maximize spatial heterogeneity and include a range of protections or management zones (5,47). These management protections included five sites within fully protected (FP) zones (otherwise known as “marine reserves”), where only non-extractive activities are permitted, three sites within general-use (GU) zones, where fishing is permitted with some gear restrictions (e.g., prohibitions on longlines, gillnets, and the use of spearguns and slings with SCUBA) and modest fishing limits (e.g., catch-size limits for queen conch and lobster), and seven sites in unprotected (NP) zones, where fishing is not restricted (46). Note that national seasonal closures for some species (e.g., Nassau grouper) and bans (e.g., on catching parrotfishes) applied to all three zones.
Benthic surveys
Benthic surveys were conducted in situ using SCUBA. At each site, dive teams laid out four to ten, 25-30 m x 2 m belt transects down the centers of reef spurs, perpendicular to the shoreline. The transects generally began on or near the shoulders of the spurs at 15–18 m depth, shoreward of the drop-off that characterizes most of the reefs, and ran upward toward the reef crest.
Transects were parallel to each other and were usually separated by > 10 m. Divers worked in buddy pairs, in which one diver laid out the transect tape and the other used a digital camera in an underwater housing to obtain videos and still-frame images of the benthos. At each site, we photographed or videotaped the belt transects at a standard distance of 25 cm above the benthos using a horizontal bar projected from the front of the camera housing. In all sampling years except 2016, we obtained underwater videos along the belt transects and extracted still frames from those videos (as outlined below). In 2016, we photographed the transects using a GoPro HERO4 by swimming at a rate of 5–7 minutes along the 30-m-long transect and taking a photograph every five seconds.
Image extraction and analysis
Because of changes in imaging technology and analytical software over the course of this study we used several techniques to extract and analyze the benthic images from the underwater transects. For sampling year 1997, we recorded Hi-8 video of each transect, using two 30-watt ultrabright lights for illumination; in 1999 and 2005, we used Sony 3chip mini DVR without illumination. From these video cassettes we randomly selected 50 frames per transect, processed the images by de-interlacing, sharpening, and enhancing them, and saved them onto a CD-ROM. In 2009, we switched to digital video. We extracted the images from the video at a rate of 1-fps using Adobe Premiere Pro CC 2014. We ran the images through the Automator program in OS-X software to select every third, fifth or seventh image, depending on the length (in time) of the transect. We analyzed 15 images/transect/site for 2009 and 2016 because we found that we could obtain a similar level of inference about community composition with 15 images per transect as with the 50 images per transect suggested by Aronson et al. (51). To select the images, we automated the process using a code in R version 3.6.3 [1] to randomly choose, copy, and paste 15 images into a new folder from our source-folder of all images.
We analyzed the benthic cover of images from 1997–2005 using Coral Point Count software (52), and from 2009 and 2016 using CoralNet (53). We manually input species-level benthic identifications for each of 10 random points overlaid onto each image (51). When species-level identifications were not possible, benthic components were identified to genus or family. All benthic components identified were pooled into five benthic categories: (1) crustose–turf–bare space (abbreviated CTB), which represents substrate that is bare, dead, covered in turf algae, and/or crustose coralline algae (48,54), (2) hard corals (which includes all scleractinian corals and milleporine fire corals), (3) macroalgae, including algae in the genus Halimeda, (4) gorgonians, and (5) sponges. The corals Orbicella annularis, O. favelota, and O. franksii were pooled as Orbicella spp. because the species complex was not divided into three distinct species during the 1997 and 1999 data collection and because they were difficult to distinguish in some video frames. In all instances, image-level point-count data were converted to percent-cover estimates, and we calculated overall mean percent covers of each category for each site and year.
Putative drivers of benthic community dynamics
We estimated local human impacts using the Global Human Influence Index (HII, version 2) from NASA’s Socioeconomic Data and Applications Center (SEDAC) database (55). The HII is a global dataset of 1-km grid cells aggregated from 1995–2004 designed to estimate location-specific human influence and thus potential impacts to natural populations and ecosystems via local direct and indirect human activities (e.g., harvesting and pollution). It is based on nine global data layers including human population density, land use, and access (which is estimated from coastlines, roads, railroads and navigable rivers). These aspects of human communities are known to be predictive of local human impacts in many natural systems including coral reefs (6,7,28,56–59). We extracted HII values for the BBR (Fig. S1) and calculated the sum of the HII scores of grid cells within a 50-km, 75-km, and 100-km buffer from the center-coordinates of each study site (Table S1). We used HII scores within the 50-km buffer for the final analysis because this metric performed well in exploratory models and it has been used successfully in prior work (5, 56). We then tested whether this index of local human impacts was related to observed changes on the monitored benthic reef communities.
Our measure of ocean-heatwave events was the site-specific frequency of Thermal Stress Anomalies (TSA Freq), obtained from the Coral Reef Temperature Anomaly Database (CoRTAD, Version 6) (60,61) (Fig. S2, Table S2). We used this metric to test for effects of thermal stress on the measured benthic groups. TSA Freq is defined as the number of deviations of 1 °C or greater from maximum weekly climatological sea-surface temperature during the 52 weeks preceding a reef survey. Other studies have found that TSA Freq is a significant predictor of coral-cover loss and coral-disease prevalence (62–64). The CoRTAD is based on 4-km-resolution weekly measurements made by the Advanced Very High-Resolution Radiometer (AVHRR) sensor (Pathfinder 5.0 and 5.2) beginning in 1982. Daytime and nighttime data were averaged weekly using data with a quality flag of 4 or better.
Data analyses
To analyze changes in benthic composition and test for the effects of potential drivers of change, we built generalized linear mixed models (GLMM) in a Bayesian setting using the blme package (65). The response variables were the logit-transformed percent covers of key benthic categories. The final models had Year, Fishing level (“fishing”, which were the sites within FP zones, and “no fishing” which included GU and NP sites), HII at the 50-km buffer, and TSA Freq as fixed effects; and Site as a random effect. A blme prior with a wishart distribution was imposed over the covariance of the random effect and modeled coefficients. All predictor variables were additive, and the REML estimation was used to fit the data as it provides unbiased estimates for the variance components. Prior to fitting models, we rescaled and centered all numerical fixed effects to optimize comparisons among variables. The final model structure for each benthic category was as follows: where α = intercept, αi = random intercept (Site), ejj = error term, and βlj – β4j are the coefficient estimates for covariates. The logit-transformed coral cover was modeled as an intercept (α), plus linear effects of Year, Fishing level, HII at 50km, and TSA Freq; a random intercept (αi) for Site, which is assumed to be normally distributed with a mean of 0 and variance σα2; and an error (eij). The index i refers to sites (i = 1, …, 15), and j refers to the year of survey (j = 1997, …, 2016). The term ejj was the within-site variance of benthic group cover and is assumed to be normally distributed with mean of 0 and a variance of σ2.
We evaluated collinearity among fixed factors by assessing variance-inflation factors and chose a threshold of 3 to determine correlated variables. We tested for homoscedasticity (equal variances across predictor variables) by plotting residuals against fitted values. Comparing fitted and residual values suggested that our models were reasonable models of the means. We also examined the marginal and conditional R-squared values of the models.
To examine changes in community composition of all benthic taxa within sites and across years, we constructed a non-metric multidimensional scaling (NMDS) ordination using the vegan package in R. We used the Bray–Curtis dissimilarity index to calculate distances among taxon-level cover data because it handles the large numbers of zeros (which denote absences) commonly found in ecological data and does not consider shared absences as being similar (66). To determine the effects of covariates (Year, TSA Freq, HII_50km, and Fishing level) on community composition changes of benthic taxa we ran a Permutational Multivariate Analysis of Variance (PERMANOVA) using the Bray-Curtis dissimilarity index to calculate distance matrices. All statistical analyses were performed in R version 3.6.3. The code and processed data are available at https://github.com/calves06/BRC.
RESULTS
Among the five benthic groups of interest—hard corals, macroalgae, CTB, gorgonians, and sponges—we identified a significant decline in hard coral and CTB cover, significant increases in macroalgal and gorgonian cover, and no change in sponge cover (Figs. 2 & S3, Table 2). Fishing status (fished versus unfished) was not predictive of observed spatiotemporal variation in hard-coral, macroalgal, CTB, or sponge cover (Figs. 2 & 3, Table 2) and was marginally and negatively related to gorgonian cover. The Human Influence Index (HII) was also unrelated to hard-coral, macroalgal CTB, or sponge cover (Fig. 3, Table 2). HII was significantly and negatively related to gorgonian cover. TSA Freq, our metric of ocean-heatwave frequency, was significantly negatively related to the cover of hard corals and gorgonians, and unrelated to the cover of macroalgae, CTB, and sponges (Fig. 3).
Throughout the two decades of this study, the substantial decline in hard-coral cover across the Belizean Barrier Reef from 26.3 % (± 7.3 SD) to 10.6 % (± 3.5 SD) (Fig. 2) was driven by a handful of reef-building coral species (Fig. 5). Notably, there was a significant decline of Orbicella spp., with mean cover falling from 12.7 % (± 7.4 SD) in 1997 to 2.2 % (± 0.9 SD) in 2016 (Fig. 5, Table S4; model estimate = - 0.719, p < 0.001). This decline was predominantly observed from 1997 to 1999, which included a major bleaching event and Hurricane Mitch (Fig. 5, Table 1), and from 2005 to 2009, which included a second bleaching event, Hurricane Dean, and an earthquake (Fig. 5, Table 1). The cover of hard-coral taxa such as Acropora spp., Colpophyllia natans, and the combined cover of “other coral” taxa (e.g., Mycetophyllia spp., Madracis spp., Favia spp. see Table S5 for a complete list) also declined significantly from 1997 to 2016 (Fig. 5, Table S4). The cover of the coral taxa Agaricia agaricities, Diploria/Pseudodiploria spp., Montastrea cavernosa, Siderastrea spp., Porites astreoides, and Porites spp. (P. porites, P. furcata, and P. divericata) remained relatively low and did not change significantly during the study period (Fig. 5, Table S4). The cover of Agaricia tenuifolia slightly but significantly increased (Fig. 5, Table S4). Fishing level and HII were not significant predictors of spatial and temporal changes of any coral taxa (Table S4), except for P. astreoides, for which sites with higher cover were associated with areas of higher HII (Table S4). The cover of Acropora spp., M. cavernosa, Orbicella spp. Porites spp., and “other coral” taxa were negatively correlated with TSA frequency (Table S4).
Based on the ordination analysis, there were major compositional shifts in the dominant benthic assemblages during 1997–2005 (left) and 2009–2016 (right) at every site (Fig. 6, Table 3), supporting the results of our models. The PERMANOVA showed that, among all covariates, time explained about 50% of the variability in benthic community changes (F = 45.8, p < 0.001) and was the only significant predictor of change in overall community composition (Fig. 6, Table 3). Fishing level, HII, and TSA frequency combined only accounted for 6% of community differences and were not good predictors of overall change of all taxa studied (Table 3). In 1997– 2005, the benthic communities of the BBR were dominated by CTB and long-lived, massive reef-building corals such as Orbicella spp. and C. natans. During 2009–2016, composition had shifted to domination by small and/or weedy hard-coral species, macroalgae, and gorgonians (Fig. 6).
DISCUSSION
Belize’s network of protected areas, designed and implemented in part to prevent the degradation of benthic reef assemblages on the BBR, has not achieved this goal. Our results complement previous findings for Belize reporting the failure of individual MPAs or the network overall to protect and restore populations of overharvested reef fishes (5,46,67,68), but see (68). We documented a statistically and ecologically significant decline in hard-coral cover, an increase in macroalgae and gorgonians, and a substantial decline of CTB, regardless of protection status (Fig. 2). Similar coral declines in isolated, well-protected, and seemingly “pristine” locations have been documented at many other sites globally (69,70).
We found that the benthic assemblages changed over time and were ecologically distinct between the earlier and later sampling intervals (1997–2005 and 2009–2016) (Fig. 6). For instance, the hard corals Acropora spp. and Orbicella spp. were more often present and more dominant (both had higher relative and absolute cover) in the early sampling years, as opposed to fleshy macroalgae and gorgonians, which dominated during later sampling years. In contrast, the cover of ‘weedy’ coral taxa such as Porites spp. and Agaricia spp. remained relatively consistent throughout the course of the study (Fig 5). The striking decline in Orbicella spp. (Fig. 5) was likely due to mortality from coral bleaching in 1998 (54,71) and 2005 (21,72–74), Hurricane Mitch in 1998, Hurricane Dean in 2007, and yellow-band disease in the early 2000s (Table 1).
Our results are concordant with previous studies in Belize that documented shifts in hard-coral and macroalgal cover (75). For example, the patch reefs of Glovers Reef atoll had ~0% hard coral and 20% fleshy-macroalgal cover in 1970–1971 but phase-shifted to 20% hard coral and 80% macroalgal cover by 1996–1997 (75). This change was due to massive declines in the reef-building corals Acropora cervicornis, A. palmata, and Orbicella spp., and large increases in the cover and biomass of fleshy and corticated seaweeds including Lobophora, Dictyota, Turbinaria, and Sargassum. Prior to the beginning of our study, acroporid abundance had already declined across much of the BBR due to both hurricanes and white-band disease (11,68,76). Most remaining A. cervicornis and A. palmata was killed by high ocean temperatures during the 1998 mass-bleaching event (54,71). A longitudinal study of A. palmata along the Mexican portion of the Mesoamerican Barrier Reef also reported declines in acroporids, with A. palmata decreasing from 7.7% in 1985 to 2.9% in 2012 (76).
We attribute changes in the benthic assemblages of coral reefs along the BBR primarily to the large-scale disturbances to the system over the last several decades, including seven hurricanes and two mass-bleaching events caused by anthropogenic climate change (Table 1). We measured the potential effects of several putative drivers, including local human impacts estimated using the Human Influence Index (HII) and the frequency of ocean heatwaves (TSA Freq). Our results indicate that the local impacts had no measurable effect on hard-coral cover. HII was, however, significantly and negatively related to changes in gorgonians and positively associated with the cover of Porites astreoides. There is abundant evidence that local impacts, including pollution, fishing, and coastal land-use practices, can severely impact coral populations (28). Yet even when these stressors are clearly present, they are often overwhelmed by the effects of large-scale disturbances including ocean heatwaves and storms (20,39,57,75).
Shifts in the dominant benthic assemblages have been documented across the Caribbean, linked to regional disturbances such as herbivore declines, coral diseases, and mass-bleaching events (2,3,69,77,78). Across seven subregions in the Caribbean, Schutte et al. (2) found significant declines in hard-coral cover and increases in macroalgal cover from 1970–2005. Corals failed to recover in the Florida Keys (79) and the U.S. Virgin Islands (80) due to subsequent, repeated disturbances. The coral reefs of Bonaire exhibited similar trends over 15 years of bleaching, storms, and diseases, with a 22% decline in coral cover and an 18% increase in macroalgal cover by 2017 (81). These trends were also apparent in our study.
The primary management response to reef degradation has been implementation of MPAs (34,41,44,82). MPAs and MPA networks are areas where extractive activities are regulated via fishing closures or gear restrictions among others. Within well-designed and enforced MPAs, fish abundance and diversity often increase and in some cases spill over into adjacent, non-protected areas (68,83–86). Some MPAs also reduce other extractive activities that could directly or indirectly impact coral populations. However, a large majority of studies have found that MPAs are not slowing or preventing the decline of reef-building corals (50,63,67,79,87–89). A recent meta-analysis of 18 studies, encompassing 66 MPAs, reported that MPAs did not affect coral loss or recovery in response to large-scale disturbances including disease, bleaching, and storms (39). Our results for the BBR agree with this broad consensus.
Unlike local human impacts, anthropogenic climate change was clearly a significant driver of the dramatic shifts in community composition that occurred on the BBR over the two-decade study. Overall coral cover, and the cover of four coral taxa—Acropora spp. Orbicella spp., Montastrea cavernosa, and Porites spp.—were negatively related to heatwave frequency (Figs. 3 and 4). This result is in agreement with other studies that have documented coral mortality and consequent declines in coral cover following the temperature-induced mass-bleaching events on the BBR in 1998 and 2005 (48,54). Many other studies have documented the role of ocean heatwaves in coral decline around the world (21,22,63,69,88–93).
Our data show a substantial shift in the state of coral reefs along the Belizean Barrier Reef over a two-decade period rife with large-scale disturbances. The results illustrate the shortcoming of protected areas in mitigating these impacts. We documented declines in the key reef-building coral genera Acropora and Orbicella, subsequent increases in macroalgal and gorgonian cover, and an overall change in the benthic assemblages over the two-decade study. Ocean-heatwave frequency was the only significant predictor of coral population declines over time. Our results provide insight into the overriding influence of regional, and global drivers at a time of rapid climate change, which will help managers improve their decision-making.
Competing interests
The authors have declared that no competing interests exist.
Data Availability
All relevant code and data are available at here: https://github.com/calves06/Belizean_Barrier_Reef_Change
Permitting
The research was performed under permits from the Belize Fisheries Department to MM, NB, KC, CF, CC, and JFB including permit numbers 000018-09 and 000028-11.
Funding
This manuscript is based upon work supported by the National Science Foundation (DGE-1650116 to CA, OCE-0940019 to JFB, and partial support from OCE-1535007 to RBA), the Rufford Small Grant Foundation, the National Geographic Society, the International Society for Reef Studies/Center for Marine Conservation Reef Ecosystem Science Fellowship, the Elsie and William Knight, Jr. Fellowship from the Department of Marine Science at the University of South Florida, the J. William Fulbright program, the Organization of American States Fellowship, the World Wildlife Fund-Education for Nature Program, the Kuzimer-Lee-Nikitine Endowment Fund, the Nicholas School International Internship Fund at Duke University, the Lazar Foundation, and the Environment, Ecology and Energy Program, the Department of Biology, and the Chancellor’s Science Scholar Research Fund at the University of North Carolina at Chapel Hill. This is contribution no. XXX from the Institute for Global Ecology at the Florida Institute of Technology. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funders.
Author contributions
MM, RBA, NB, CF, and JFB designed the study
MM, RBA, NB, CC, CF, and JFB obtained the study funding
MM, NB, KC, CC, CF, CA, LM, AV, and JFB performed the surveys
MM, NB, CF, CC, and JFB performed the pre-survey expedition planning
CA, MM, NB, CF, CC, and AV organized the study data
MM, NB, CF, CA, ZL, LM, and JFB analyzed the benthic videos and images
CA, JU, AV, and JFB analyzed the data
CA, RA, AV, and JFB wrote the manuscript with contributions from the other authors
Acknowledgements
We thank the many volunteers who assisted with data collection, logistical support and image analysis over the years. We are extremely grateful for our partnerships with staff from the Belize Fisheries Department, Belize Coastal Zone Management Project, Wildlife Conservation Society, Marine Research Center at the University of California–Berkeley, Belize Audubon Society, Pelican Beach Resort, Rum Point Inn, Sea Sports Belize, Healthy Reefs for Healthy People, Bacalar Chico National Park and Marine Reserve, Hol Chan Marine Reserve, The Nature Conservancy, Southern Environmental Association, Toledo Institute for Development and Environment and the Smithsonian Institution.
LITERATURE CITED
- 1.↵
- 2.↵
- 3.↵
- 4.
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.
- 24.
- 25.
- 26.
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.
- 34.↵
- 35.
- 36.
- 37.↵
- 38.↵
- 39.↵
- 40.
- 41.↵
- 42.↵
- 43.
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.
- 85.
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.
- 91.
- 92.
- 93.↵
- 94.
- 95.
- 96.
- 97.