- Email: [email protected]
Spatial and temporal co-structure analyses between ichthyofauna and environment: an example in the tropics
C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646 © 2001 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS. Tous droits réservés S0764446901013385/FLA
Ecology / Écologie
Spatial and temporal co-structure analyses between ichthyofauna and environment: an example in the tropics Laurence Blanc, Catherine Aliaume*, Alfonso Zerbi, Gérard Lasserre Université Montpellier II, laboratoire hydrobiologie marine et continentale, CNRS UMR 5556, CC 093, place E.-Bataillon, 34095 Montpellier cedex 05, France Received 20 October 2000; accepted 12 March 2001 Communicated by Claude Combes
Abstract – Ichthyofauna distribution and habitat characteristics of Thalassia beds in the Grand Cul-de-Sac marin lagoon in Guadeloupe were studied during a one-year survey. Environmental variables (9) were measured monthly in ten sites along with collection of fish communities. The environmental data set, analysed alone through between–within group ‘principal component analysis’ (PCA), exhibited a significant spatial and temporal variability. The fish data set, however, presented only a significant spatial structure, stable over the year. Given the lack of temporal variability in fish distribution, a ‘between-site co-structure analysis’ (BSCA) was used to compare the faunistic and environmental structures in space. The co-inertia structure was reduced to one axis representing a strong coast-reef gradient, the major common phenomena to both data sets. Environment and fish distribution allowed to distinguish sites directly under mangrove influence (characterised by high seagrasses, high concentration of chlorophyll a and high densities of zooplankton), to sites under reef influence (with short but dense seagrasses, clear water, and poor nutriments). For that purposes, the BSCA summarised efficiently what in common the fauna spatial structure and the environment spatial structure may present. © 2001 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS fish community / species-environment relationships / co-inertia analysis / seagrass / Guadeloupe
Résumé – Analyse de la co-structure spatiale et temporelle ichtyofaune–environnement : un exemple en milieu tropical. La distribution de l’ichtyofaune et les caractéristiques environnementales de dix sites d’herbiers à Thalassia du Grand Cul-de-Sac Marin en Guadeloupe ont été étudiées mensuellement pendant 12 mois. Le fichier des variables environnementales, traité séparément par analyse en composantes principales (ACP) inter–intragroupes, a présenté une variabilité spatiale et temporelle significative. En revanche, le fichier des biomasses par espèces de poissons a présenté seulement une variabilité spatiale significative, stable dans le temps. Étant donné le manque de variabilité temporelle pour les poissons, seule une analyse de co-structure inter-station (ACIS) a été tentée entre les deux jeux de données. Les résultats de l’analyse de co-inertie spatiale ont permis de mettre en évidence un axe fort, commun aux deux tableaux, qui représente un gradient côte–récif, avec des sites sous l’influence de la mangrove colonisés d’herbiers de grande taille, de fortes concentration de chlorophylle a et de zooplancton, opposés aux sites davantage sous l’influence de récifs coralliens
*Correspondence and reprints. E-mail address: [email protected] (C. Aliaume).
635
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
avec des herbiers plus courts mais denses, une eau plus claire mais pauvre en nutriments. Dans le cas présenté, l’analyse de co-inertie synthétise efficacement ce que la structure spatiale des communautés de poissons peut avoir en commun avec la structure spatiale de l’environnement. © 2001 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS communauté de poissons / relations espèces–environnement / analyse de co-inertie / herbiers / Guadeloupe
Version abrégée
Dans le but d’étudier l’ichtyofaune des herbiers à Thalassia du Grand cul-de-sac Marin (Guadeloupe) et ses relations avec les descripteurs de l’environnement, dix stations ont été échantillonnées mensuellement de décembre 1987 à novembre 1988, simultanément aux mesures de sept descripteurs de l’environnement : densité de plants de Thalassia, longueur moyenne de feuille, salinité, transparence, température de l’eau, concentration en nitrates, silicates, chlorophylle a, et densité en copépodes. Dans un premier temps, l’analyse spatiale et/ou temporelle des données a été réalisée par ANOVA à deux facteurs, puis complétée par ACP inter- et intraclasses (dates ou stations). Les pourcentages de variabilité inter-classes obtenus sont ensuite jugés significatifs ou non par des tests de permutations. De cette étude préalable des variables, on conclut que le milieu n’est ni stable dans le temps, ni homogène dans l’espace. Transparence, température, salinité et chlorophylle a varient significativement (p < 0,01) dans l’espace et dans le temps. Densité en copépodes, longueur et densité des herbiers présentent une variabilité spatiale significative (p < 0,01). Seuls les nitrates et les phosphates ne présentent pas de variabilité significative ni dans l’espace ni dans le temps. L’ACP inter-dates des données environnementales oppose le long de son premier axe factoriel (68 %) les mois de la saison sèche aux mois de la saison humide, avec la température et la salinité pour descripteurs saisonniers. L’analyse inter-stations permet d’opposer le long du premier axe (65 %) les stations du large aux stations côtières grâce essentiellement aux descripteurs de struc-
1. Introduction Coastal aquatic ecosystems (such as lagoons, estuaries, creeks or marsh) are inhabited by marine, fresh-water or brackish fish populations which exhibit various habitat use strategies: they may be migrant or resident, using the habitat as juvenile or adult, for reproduction or foraging ground. Most of these ecosystems play an important role as nurseries for numerous species of fish, offering food and shelter to the larval and juvenile stages [1–6]. It is also well documented that fish species are distributed depending on
636
ture des herbiers et aux concentrations de chlorophylle a et copépodes. Sur le second axe (19 %) apparaît essentiellement un gradient de salinité. L’ANOVA à deux facteurs réalisée sur les espèces de poissons a conclu à un effet mois significatif dans 15 % des cas et à un effet station significatif dans plus de 80 % des cas. La faune n’est donc pas homogène dans l’espace, elle est en revanche plutôt stable dans le temps. Lors de l’ACP inter-date, la structure spatiale faunistique a fait ressortir un gradient côte–récif très net. Dans les stations plus au large, des espèces récifales sont mélangées à des espèces plus typiques d’herbiers. Dans les stations côtières, on a pu distinguer un peuplement sous l’influence plus marquée de la mangrove, d’un peuplement à double tendance récifale et mangrove que l’on peut qualifier d’intermédiaire. Cohérente avec les analyses séparées, l’analyse de co-structure spatiale a mis en évidence un gradient côte–récif assez marqué, permettant de distinguer les stations directement sous l’influence de la mangrove caractérisées par des herbiers longs et de fortes concentrations en chlorophylle a et copépodes, des stations sous influence récifale caractérisées par une forte transparence des eaux et des herbiers denses. Le gradient côte–récif représente le principal phénomène commun aux deux tableaux de données (un seul axe représentant 93 % de l’inertie initiale). Le gradient de salinité qui jouait un rôle dans la structure spatiale du milieu, n’apparaît plus dans la co-structure spatiale révélant ainsi que la distribution spatiale des espèces de poissons n’est pas liée aux variations spatiales de ce facteur.
environmental gradients, life cycles and species interactions. Among the environmental descriptors showing a significant effect in fish community structure of coastal systems we may cite: temperature [7–10], salinity [8, 9, 11–13], dissolved oxygen [14, 15], water depth [16–18], turbidity [2, 19, 20], morphological characteristics of the substrate [21–23]. Our understanding of the relationships between habitat and fauna distribution is a corner stone in determining habitat value and productivity potential. The present study deals with the fish community associated with seagrass ecosystem of the Grand Cul-de-Sac
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
marin, the largest lagoon of Guadeloupe (French West Indies). Seagrasses, largely composed of Thalassia testudinum Banks ex Konig, are known to serve as nursery ground for numerous fish species [24–27]. It serves also as a passage way for migrating species [28] and may be used as direct food source by herbivorous species [29–31]. If fish in mangrove [32–35] and in coral reefs [36–38] have been studied for decades in Guadeloupe, it is only recently that scientific interest has focused on fish communities inhabiting seagrass beds [25, 39–41]. In this study, we characterize fish assemblages associated with the Thalassia beds in space and time and we hypothesize that there should be some agreement between the ichthyofauna structure and the environment structure.
As a consequence the objectives of this study are 1) to determine spatial and temporal structure of fauna and habitat data sets in a separate analysis, and 2) to demonstrate the agreement between faunistic and environmental data.
2. Materials and methods 2.1. Data and site description
The Grand Cul-de-Sac marin lagoon of Guadeloupe island is a large bay (11 000 ha) limited seaward by a 30-km-long barrier reef and landward by an important mangrove forest (figure 1). Seagrasses have settled in this
Figure 1. Sampling location of the Grand-Cul-de-Sac-Marin (Guadeloupe, FWI)
637
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
differences in space and time of the environmental and biological factors. To comply with normality assumption, simple normalising transformations (Log (x), X0,5) were performed on environmental variables.
well protected area and usually occupy habitats from very shallow water (less than 0.5 m deep) to 3 m of depth. During a stratified pre-sampling survey [41], seagrasses (Thalassia) were studied (length, densities and biomass recorded) on 20 sites and analysed to determine stations with redundant characteristics. Some sites had to be eliminated to comply with fishing gear constraints (depth limited to 1.5 m, no presence of corals). As a result, ten sites, where seagrass characteristics showed significant differences, were selected (figure 1): three sites were adjacent to a coral reef barrier (I, J, M), four sites were near a mangrove fringe (N, P, Q, R), one site was near a river (C) and two sites were inside the lagoon (A, B). All sites were sampled monthly from December 1987 to November 1988. Fish fauna was collected using a bag seine (50 m long, 2 m high with a square mesh ranging from 10 mm at the aisle to 3 mm in the bag). Three hauls were pulled at each site since this effort was demonstrated to be the best compromise between sampling cost and fauna representativity [39]. To eliminate the probable lunar influence, samples were always collected in the morning during the last quarter phase, which corresponded to the maximum sampling vulnerability [42]. The final fauna data set showed species biomass collected in 120 samples. One hundred species were identified, however, only species representing at least 0.1 % of the total biomass and present at least in 5 % of samples (38 species) were kept in the present analyses to avoid bias associated with rare species. Environmental factors chosen for their potential direct or indirect effect on fish distribution were: Thalassia blade density (number of blades per m2 – DENS); blade length (cm – LENG); transparency, coded from 1 to 12 with a Secchi disk (TRAN); surface temperature (taken 30 cm below surface – °C – TEMP); surface salinity (taken 30 cm below surface – SAL); nitrate concentration (mg·L–1 – NO3); silicate concentration (mg·L–1 – SiO2); chlorophyll a (mg·m–3 – CLA); copepod density (numbers·m–3 – COP), representing 90 % of the zooplankton density. These descriptors were measured simultaneously to the fish collection at all sites, except for the seagrass density and length, which were measured only at the beginning of the survey and supposed to remain stable over the year of the study.
Multivariate analyses such as between-group / withingroup ‘principal component analyses’ [44, 45] use time and space as instrumental variables that control the analysis. Between-group analysis focuses on the difference between groups (e.g. between sites); it seeks for axes that will discriminate best the centres of gravity of each group. For example, in our study, the between-site analysis of the fauna (X1, 38 species, 120 samples) will be a PCA of the site mean set (X1m, 38 species, 10 site means). In order to highlight the variability of each sample around the group centre, the initial data of table I are used as supplementary individuals. Within-group analysis seeks for axes shared by the groups. Therefore the within-site analysis of the fauna will be a PCA of the data set centred by site (X1–, 38 species, 120 centred by site values). The statistical significance of between-group analysis is tested using a MonteCarlo permutation test [46]. To answer the question of whether there is an agreement between the structure defined by the fauna and the structure defined by the environment one can use the co-inertia analysis. This analysis enables to examine species–environment relationships when many species and environmental variables are sampled in a small number of sites. The agreement between the two data sets is evaluated by determining the co-structure. The method calculates factorial axes by maximising the covariance between the factorial scores of the two data sets [47]. So, the analysis maximises the variance, which defines the structure of each table separately but also, the correlation between the two new sets of projected co-ordinates. The advantage of co-inertia analysis is demonstrated by its extensive use in freshwater biology [48–55]. Moreover between-site analysis and co-inertia analysis may be combined to study the spatial co-structure between species and their environment [56, 49]. This between-site co-structure analysis (BSCA) consists then in finding out a combination in each table of average values maximising the covariance among the between-site environmental axes and the between-site faunistic axes. To do so we first reduced the two data sets to their site mean sets, and then calculated the vectorial covariance between environmental and ichthyological variables. Finally we performed a PCA on this new matrix to define new factorial axes.
2.2. Data processing
Prior to the data processing fish biomass was transformed using double square-root to prevent abundant species from dominating the analysis as advised by Field et al. [43]. A preliminary two-way ANOVA was used to test
Table I. Environmental characteristics. N = 120 (12 months × 10 sites). TRAN = transparency; TEMP = temperature; SAL = salinity; NO3 = nitrate; SIO2 = silicate; CLA = chlorophyll a; COP = copepod density; DENS = number of blades m–2; LENG = seagrass length.
Factor mean standard-deviation
638
TRAN
TEMP (°C)
SAL
NO3 (mg·L–1)
SIO2 (mg·L–1)
CLA (mg·m–3)
COP (m–3)
DENS (m–2)
LENG (cm)
6.7 2.7
29.1 1.5
35.4 1.9
1.6 0.3
2.2 1.5
0.9 0.8
50.3 47.4
454 150.6
27.7 7.5
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
Data treatments were performed using ADE-4 [57], a free software package available on the internet (http://pbil.univ-lyon1.fr/ADE-4/).
3. Results 3.1. Spatial and temporal analyses of the environmental data set
Means and standard deviations of the environmental factors, all sites and months included, are displayed in table I. Temporal variation of the 9 environmental factors (normalised and, for some, transformed data) are plotted by site (figure 2). The length and the density of Thalassia are constant through time since the measure was not repeated and show only spatial variability. Temperature and salinity show a seasonal pattern with lower temperature and higher salinity during the dry season (December to May), and inverse tendencies during the rainy season (June to November). An important drop of the salinity is observed in site C, due to the influence on the nearby river (Goyave River, the largest river of Guadeloupe). Transparency shows a marked spatial pattern. For the four other factors no clear pattern can be seen. However, one may notice a strong increase of SiO2 concentrations in 3 sites (C, M, N) during May and a strong increase in chlorophyll a in all sites (but specially in B, N, P, Q et R) during February. The two-way ANOVA results are given in table II. There is no significant difference between sites or between months for NO3 and SiO2, which will lead us to eliminate these two variables in the rest of the analyses. There is a significant spatial difference between the copepod density, the length and the density of Thalassia. All the other environmental variables show a significant difference in both time and space. These preliminary results indicate that the environment is not stable in time and not spatially homogeneous. Between–within-group analyses were performed for space and time effect on the environmental data set (table III). The total inertia may be split into between-date inertia (representing 25 %) and within-date inertia (75 %).
Figure 2. Monthly variation of normalised environmental variables, each curve represents one site. TRAN = Transparency; TEMP = Temperature ( °C); SAL = Salinity; NO3 = Nitrate (mg·L–1); SIO2 = Silicate (mg·L–1); CLA = Chlorophyll a (mg·m–3); COP = Copepod density (m–3); DENS = number of blades m–2; LENG = Seagrass length (cm).
Table II. Two-way ANOVA performed on transformed data of environmental variables. TRAN = transparency; TEMP = temperature; SAL = salinity; NO3 = nitrate; SIO2 = silicate; CLA = chlorophyll a; COP = copepod density; DENS = number of blades m–2; LENG = seagrass length; ns: non-significant at p < 0.01; n = 120 (12 months × 10 sites); df between sites = 9; df between month = 11.
Factor Month effect Site effect
TRAN ns p < 10–4
TEMP p < 10 ns
–4
SAL p < 10 p = 3.10–2 –4
NO3 ns ns
SIO2 ns ns
CLA –4
p = 3.10 p < 10–8
COP
DENS
LENG
ns p < 10–7
p < 10–4
p < 10–4
Table III. Inertia and first eigen value (λ1) associated with global, between- and within-group PCAs of the environment data (np : not performed due to time constant factors).
PCA
global
within-date
between-date
within-site
between-site
Inertia λ1
7 2.62
5.2 (75 %) 2.53
1.8 (25 %) 1.21
3.4 (49 %) np
3.6 (51 %) 2.29
639
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
It may also be split into between-site inertia (51 %) and within-site inertia (49 %). The within-site PCA was not performed due to constant values of environmental factors (blade lengths and densities). Permutation tests for both between-date and between-site structure are highly significant (P < 0.0001), therefore both analyses are justified. In the between-date analysis (figure 3), the first factorial axis, representing 68 % of the inertia (figure 3A), opposes dry season months (December to May) to wet season months (June to November) (figure 3C). The two environmental factors exhibiting a clear seasonal pattern are salinity and temperature which are highly correlated with the first factorial axis (figure 3B). A much lesser phenomena is expressed by the second axis (18 %) which mostly opposes February to March, chlorophyll a to copepod. The between-site analysis (figure 4) present similar explanation capacity (65 % and 19 % for the first two axes
Figure 4. Between-site PCA of the environmental data set. A: Eigen values. B: 1–2 factorial map of the variables. C: Projection of the group means (circles) linked to their corresponding samples projected as supplementary data (small squares). TRAN = Transparency; TEMP = Temperature (°C); SAL = Salinity; NO3 = Nitrate (mg·L–1); SIO2 = Silicate (mg·L–1); CLA = Chlorophyll a (mg·m–3); COP = Copepod density (m–3); DENS = number of blades m–2; LENG = Seagrass length (cm).
Figure 3. Between-date PCA of the environmental data set. A: Eigen values. B: 1–2 factorial map of the variables. C: Projection of the group means (circles) linked to their corresponding samples projected as supplementary data (small squares). Months are identified by # 1 for January… TRAN = Transparency; TEMP = Temperature (°C); SAL = Salinity; NO3 = Nitrate (mg L–1); SIO2 = Silicate (mg·L–1); CLA = Chlorophyll a (mg·m–3); COP = Copepod density (m–3); DENS = number of blades m–2; LENG = Seagrass length (cm).
640
respectively – figure 4A). Sites are ranged along the first axis following more or less a coast–reef gradient (figure 4C). Sites A, N, P, and R show similar environmental characteristics, with long blades and rather low density of seagrasses, high concentrations of chlorophyll a and copepods, and low transparency. Sites I and M have very similar habitat trends, with high transparency, short but dense seagrasses, low concentrations of chlorophyll a and copepods. Sites B, C, and Q show intermediate tendencies. According to its environmental characteristics, site J is more associated with coastal sites than seaward sites, and it is primarily due to its medium size seagrass and medium range transparency. Along the second axis, site C is opposed to J due to salinity gradient and seagrass densities.
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
3.2. Spatial and temporal analyses of the ichthyofauna data set
Compared to the environment, fish community does not show a significant temporal variation.
The 38 species used in the analyses are listed in table IV, along with their relative percentage in biomass contribution, their occurrence pattern and their preferred habitat as adult [58]. The results of the two-way ANOVA performed on the 38 species separately showed that differences between sites were significant (p < 0.01) for all species except seven (Hemiramphus brasiliensis, Caranx latus, Lutjanus griseus, Bairdiella ronchus, Scorpaena grandicornis, Sparisoma chrysopterum, and Acanthurus bahianus). However, only six species (Haemulon plumieri, Haemulon sciurus, Pseudupeneus maculatus, Eucinostomus gula, Chaetodon capistratus and Sphaeroides spengleri) show a significant temporal difference (p < 0.01). The ichthyofauna will be therefore considered stable in time but not in space.
Like for the environmental data between–within-group analyses were performed on the ichthyofauna data (table V). The global PCA and the within-date analyses are very similar since there is no temporal effect. The betweendate analysis is therefore not necessary. However, the between-site analysis, representing 38.5 % of the total inertia, is highly significant (Monte Carlo permutation test – P < 0.001). The first and the second factorial axes represent respectively, 58 % and 14 % of the between-site inertia (figure 5A). The factorial map of site projection (figure 5C) shows similar trends to the one issued from the environmental analysis particularly the proximity of sites I-M, A-N-P and B-Q-R, and the coast-reef gradient given by the first axis. Along axis 2, site J is opposed to its two
Table IV. List of fish species and their codes for graphs. [1] Relative percent of biomass contribution, [2] Occurrence status in the area (R = collected in less than 50 % of the months; C = collected in 50 to 80 % of the months; P = collected in more than 80 % of the months), [3] Preferred habitat as adult (R = Coral reefs; S = Seagrass beds; M = Mangrove; P = Pelagic), [4] Results of the 2-way ANOVA : month effect (ns = not significant p = 0.01; * = significant p < 0.01; df = 11), [5] Results of the 2-way ANOVA : site effect (ns = not significant p = 0.01; * = significant p < 0.01; df = 9),
Code
species
Hja Aly Hbr Ast Sba Hpu Sfl Cla Lan Lap Lgr Och Hbo Hch Hfl Hpl Hsc Arh Cbo Bro Pma Ear Egu Sco Cca Ele Lma Scr Sch Sra Svi Aba Ach Mci Sgr Ssp Ste Dho
Harengula jaguana Anchoa lyolepis Hemiramphus brasiliensis Atherinomorus stipes Sphyraena barracuda Hypoplectrus puella Serranus flaviventris Caranx latus Lutjanus analis Lutjanus apodus Lutjanus griseus Ocyurus chrysurus Haemulon bonariense Haemulon chrysargyreum Haemulon flavolineatum Haemulon plumieri Haemulon sciurus Archosargus rhomboidalis Calamus bojanado Bairdiella ronchus Pseudupeneus maculatus Eucinostomus argenteus Eucinostomus gula Scorpaena grandicornis Chaetodon capistratus Eupomacentrus leucostictus Lachnolaimus maximus Scarus croicensis Sparisoma chrysopterum Sparisoma radians Sparisoma viride Acanthurus bahianus Acanthurus chirurgus Monacanthus ciliatus Sphaeroides greeleyi Sphaeroides spengleri Sphaeroides testudineus Diodon holacanthus
[1]
[2]
[3]
[4]
[5]
0.6 4.4 2.1 1.0 4.9 0.9 0.9 0.2 0.2 1.0 0.2 32.4 0.6 0.3 1.1 3.1 0.6 2.2 0.3 1.0 0.4 0.3 16.4 0.1 1.5 0.7 0.1 1.7 0.7 0.2 0.1 0.3 0.2 0.5 0.1 0.9 0.8 14.8
P P C P P P P C C P P P P P P P P P C R C P P R P P C P P P P P P P P P P P
P M M/P S/R R/P/S S/R S R/P R R/M M/R S/R R/S/M R R R R M/S R M R M M S R R R R R/S S R R R S M S M M/R
ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns * * ns ns ns * ns * ns * ns ns ns ns ns ns ns ns ns ns * ns ns
* * ns * * * * ns * * ns * * * * * * * * ns * * * ns * * * * ns * * ns * * * * * *
641
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
Table V. Inertia and first eigen value (λ1) associated with global. Between- and within-group PCAs of the faunistic data.
PCA
global
within-date
between-date
within-site
between-site
Inertia λ1
41 10.51
37.3 (91 %) 10.21
3.7 (9 %) ns test
25.2 (61 %) 3.29
15.8 (39 %) 9.13
neighbouring sites I-M, and sites P-A are opposed to sites B-Q-R, which seems to reflect differences between exposed (I, M, B, Q, R) to protected (J, P, A) sites. Figure 5B (projection of the species scores) allows to characterise
site groups according to fish species composition (only species contributing to 1 % of the first axis and 1 % of the second axis are mentioned here): – sites I, J, M, are characterised by reef species (Sparisoma croicensis, Eupomacentrus leucostictus, Sparisoma radians, Pseudupeneus maculatus) mixed with seagrass species (Monacanthus ciliatus, Sparisoma viride) – sites A, N, P which are under direct influence of mangrove, present mangrove species (Eucinostomus gula, Sphaeroides testudineus, Sphaeroides greeleyi, Eucinostomus argenteus, Diodon holocanthus, Archosargus rhomboidalis, Anchoa lyolepis), mixed seagrass species (Serranus flaviventris, Hypoplectrus puella), and reef species known to spend juvenile phase in seagrasses (Haemulon bonariense, Lutjanus analis, Lachnolaimus maximus) – intermediate sites B, C, Q, R, colonised by a mix of reef species (Chaetodon capistratus, Haemulon chrysargyreum, Acanthurus bahianus), mangrove species (Lutjanus apodus, Hemiramphus brasiliensis), and widely spread species (Ocyurus chrysurus). Sphyraena barracuda and Haemulon plumieri are particular species specially represented in more protected areas like A, P and J. To help visualising species distribution, scores of gravity centres are plotted (figure 6A) along with values of biomass centred over 12 months for selected species (figure 6B). This figure shows how some species, along axis 1, colonise clearly along the shore (Eucinostomus gula) or near the barrier reefs (Scarus croicencis, Monacanthus ciliatus). 3.3. Spatial co-structure between fauna and environment
Figure 5. Between-site PCA of the faunistic data set. A: Eigen values. B: 1-2 factorial map of the species. C: Projection of the site means (circles) linked to their corresponding samples projected as supplementary data (small squares).
642
Since the fauna data set did not show a significant temporal difference, the co-inertia analysis is limited to between-site co-structure of faunistic and environmental data to study the spatial structure shared by both matrices. As expected from the separate analyses, there is a significant between-site co-structure of species and environment factors (Monte Carlo permutation test, P < 0.001). The co-structure is mainly one-dimension (F1 representing 93 % of the co-structure inertia), and score projections will be therefore on a single axis (figure 7). This high correlation (0.93) between faunistic and environmental scores on this first co-structure axis confirms the strong spatial co-structure between the two data sets. Environmental scores projected on the co-inertia axis (figure 7B) confirms the main gradient observed in the environmentalone analysis with the opposition of blade length, chlorophyll a and copepods to blade density and water transparency. Likewise, fauna scores (figure 7C) show the same trends in species distribution; with coral reef species
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
however its environmental characteristics are similar to those found in intermediate sites like B, C and Q. Site Q shows also a disagreement, its fauna is similar to the one of site R but its habitat characteristics are closer to sites B or J. The differences in site I is barely significant and more difficult to interpret. For the seven other sites the difference is not significant and they all show a good agreement between fauna and environment. It is to be noticed that extreme sites such as I, M, A, N and P exhibit less variable environment or fauna structure as compared to intermediate sites such as B, Q and C more exposed to either environment variability or fish movements.
4. Discussion
Figure 6. Ichthyofauna mapping based on the between-site PCA results. A: F1 (left) and F2 (right) score mapping of centres of gravity. B: Score mapping of biomass centred over 12 months for selected species. The surface of the circle (positive value) and of the square (negative value) is proportional to the absolute value. See for the meaning of species codes.
opposed to coastal species. The distance between the centres of gravity (white circles for faunistic data and black circles for environmental data – figure 7D) gives an estimate of the agreement between fauna and environment. A shorter distance represents a better agreement. For sites J (P < 0.0001), Q (P < 0.001), and J (P < 0.05), the difference in score means between fauna and environment is significant (Wilcoxon rank test, n = 12). Site J presents a fauna composition close to those found in sites I and M
The separate between-group analyses allow independent analyses on spatial and temporal variation of a data set (either fauna or environment). In our study such analyses led to define spatial structures for both fish and habitat data sets. The environment showed a prominent coast-reef gradient expressed by seagrass morphological structures (higher in calm waters but also less dense in muddy substrate of the coastal sites, short in agitated waters of seaward sites), and productivity levels (higher concentration of chlorophyll a and density of copepods towards the coast). Site J, even though in a seaward area, is not characteristic of his area since it is a sheltered site and presents relatively calm and turbid waters. Seasonal variations are clearly demonstrated for environmental variables such as temperature and salinity, which follow the usual pattern of rainfall in the Caribbean region. Other temporal changes, this time more drastic, may be observed for variables such as chlorophyll a and densities in zooplankton which follow the episodic pattern of blooms. The fauna distribution shows a clear spatial pattern with sites under reef influence (IJM) characterised by parrotfish (Sparisoma croicensis, S. radians, S. viride), goatfish (Pseudupeneus maculatus) and damselfish (Eupomacentrus leucostictus), and sites under mangrove influence (ANP) with fauna dominated by moharras (Eucinostomus gula), puffers (Sphaeroides testudineus, S. greeleyi), grunts (Haemulon bonariense), and seabream (Archosargus rhomboidalis). These species are well established in their habitat and present very little temporal variation over the year. They share their habitat with seasonal species like grunts (Haemulon plumieri, H. sciurus) and butterflyfish (Chaetodon capistratus). This temporal stability despite varying environmental conditions, suggests that the community structure in the Grand Cul-de-Sac marin lagoon is in ‘equilibrium’ state [59], regulated more by biological interactions, predation or resource-sharing than by environmental fluctuations. Baelde [25] explained fish community stability in Guadeloupe seagrass beds by the continuous migration of fish from coral reefs to seagrasses. Louis et al. [14] observed a relative temporal fish community stability in mangrove ecosystem when studying individual abundance or biomass. Other studies demonstrated fish stability based on the survey of species composition or
643
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
Figure 7. Between-site co-structure analysis of fauna-environment data sets. A: Eigen values. B: Environment score projection. C: Fauna score projection. D: Site mean score projection seen through the fauna (white circles) and through the environment (black circles) link to their corresponding samples.
distribution in coral reefs [60–63], or based on species richness in temperate reefs [64]. On the other hand, few studies demonstrated temporal variations of fish communities in tropical lagoons [65], in coral patch reefs [66] or in mangrove areas [9]. It is to be noticed that the present study did not take into account the biodiversity and the ephemeral entrance of rare species since we excluded from the analyses more than 60 species (biomass contribution < 0.1 % of the total catch). We voluntarily used the more abundant species to draw the general patterns of the fish communities in seagrass beds. The data collection was also limited to one year of observations which does not allow inter-annual comparisons. While temporal stability in fish community of tropical ecosystems may not be commonly observed, a spatial structure however has been demonstrated in many studies [14, 25, 60–62, 66, 67]. In our study, as both environmental and fish data showed a coherent spatial distribution, a
644
spatial co-inertia analysis was justified and expected to express what the two data sets had in common spatially. The overall variation explained in between-site co-inertia was reduced to one axis (representing 93 % of it), whereas the separate between-site analyses led to first axes explaining 65 % and 58 % of the initial inertia (for the fauna set and the habitat set respectively). This is explained by the fact that the major phenomena common to both data sets is the coast–reef gradient and that most of the sites show a good agreement between environmental characteristics and ichthyofauna composition. The environmental variables involved in that axis such as seagrass height and density, and nutriments present a coherent structure with species distribution and may be involved in structuring the fish community. Water turbidity and vegetation complexity are known to play an important role in predation escapement and juvenile survival [2, 19, 27]. The other phenomena showed in separate analyses by the second
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
axis (salinity gradient for the habitat and ‘shelter’ gradient for the fauna) disappeared in the co-inertia analysis. This revealed that the spatial distribution of fish species in this study is not linked to the spatial variation of salinity. Salinity influence on the fish community has in fact been discussed and in some instances, species do not show preferences to salinity gradients and may be influenced to some other more important factors [68] such as water turbidity [69], dissolved oxygen [14] or habitat complexity [23]. In conclusion, the between-site co-inertia analysis
summarises efficiently what in common the fauna structure and the environment structure may present spatially, regardless of the temporal aspect.
References
[19] Cyrus D.P., Blaber S.J.M., The influence of turbidity on juvenile marine fishes in estuaries. Part 1. Field studies at Lake St. Lucia on the Southeastern coast of Africa, Exp. Mar. Biol. Ecol. 109 (1985) 53–70. [20] Bottom D.L., Jones K.K., Species composition, distribution, and invertebrate prey of fish assemblages in the Columbia River Estuary, Prog. Oceanogr. 25 (1990) 243–270. [21] Stoner A.W., Distribution of fishes in seagrass meadows: role of macrophyte biomass and species composition, Fish. B. 81 (1983) 847–856. [22] Jenkins G.P., Wheatley M.J., The influence of habitat structure on nearshore fish assemblages in a southern Australian embayment: Comparison of shallow seagrass, reef-algal and unvegetated sand habitat, with emphasis on their importance to recruitment, J. Exp. Mar. Biol. Ecol. 221 (1998) 147–172. [23] Garcia-Charton J.A., Perez-Rufaza A., Correlation between habitat structure and a rocky reef fish assemblages in the southwest Mediterranean, Mar. Ecol. 19 (1998) 111–128. [24] Livinsgton R.J., Trophic organization of fishes in a coastal seagrass system, Mar. Ecol.-Prog. Ser. 7 (1982) 1–12. [25] Baelde P., Differences in the structures of fish assemblages in Thalassia testudinum beds in Guadeloupe, French West Indies, and their ecological significance, Mar. Biol. 105 (1990) 163–173. [26] Rooker J.R., Holt S.A., Utilization of subtropical seagrass meadows by newly settled red drum Sciaenops ocellatus: Patterns of distribution and growth, Mar. Ecol.-Prog. Ser. 158 (1997) 139–149.
[1] Weinstein M.P., Shallow marsh habitats as primary nurseries for fishes and shellfish, Cape Fear River, North Carolina, Fish. B. 77 (1979) 339–357. [2] Blaber S.J.M., Blaber T.G., Factors affecting the distribution of juvenile estuarine and inshore fish, J. Fish Biol. 17 (1980) 143–162. [3] Weinstein M.P., Weiss S.L., Walters M.F., Multiple determinants of community structure in shallow marsh habitats Cape Fear River estuary, Noth Carolina, U.S.A, Mar. Biol. 58 (1980) 227–243. [4] Loneragan N.R., Potter I.C., Lenanton R.C.J., Caputi N., Spatial and seasonal differences in the fish fauna in the shallows of a large Australian estuary, Mar. Biol. 92 (1986) 575–586. [5] Tzeng W.N., Wang Y.T., Structure, composition and seasonal dynamics of the larval and juvenile fish community in the mangrove estuary of Tanshui River, Taiwan, Mar. Biol. 113 (1992) 481–490. [6] Yanez-Arancibia A., Lara-Domiguez A.L., Day J.W. Jr., Interactions between mangrove and seagrass habitats mediated by estuarine nekton assemblages: Coupling of primary and secondary production, Hydrobiologia 264 (1993) 1–12. [7] Morin B., Hudon C., Whoriskey F.G., Environmental influences on seasonal distribution of coastal and estuarine fish assemblages at Wemindji, eastern James Bay, Environ. Biol. Fish. 35 (1992) 219–229. [8] Rogers S.I., Millner R.S., Factors affecting the annual abundance and regional distribution of English inshore demersal fish populations: 1973 to 1995, ICES J. Mar. Sci. 53 (1996) 1094–1112. [9] Laroche J., Baran E., Rasoanandrasana N.B., Temporal patterns in a fish assemblage of a semiarid mangrove zone in Madagascar, J. Fish Biol. 51 (1997) 3–20. [10] Tzeng W.N., Wang Y.T., Chern Y.T., Species composition and distribution of fish larvae in Yenliao Bay, Northeastern Taiwan, Zool. Stud. 36 (1997) 146–158. [11] Rozas L.P., Odum W.E., Use of tidal marshes by fishes and macrofaunal crustaceans along a marsh stream-order gradient, Estuaries 10 (1987) 36–43. [12] Odum W.E., Comparative ecology of tidal freshwater and saltmarshes, Ann. Rev. Syst. 19 (1988) 147–176. [13] Thiel R., Sepulveda A., Kafemann R., Nellen W., Environmental factors as forces structuring the fish community of the Elbe Estuary, J. Fish Biol. 46 (1995) 47–69. [14] Louis M., Bouchon C., Bouchon-Navaro Y., Spatial and temporal variations of mangrove fish assemblages in Martinique (French West Indies), Hydrobiologia 295 (1995) 275–284. [15] Deegan L.A., Finn J.T., Ayvazian S.G., Ryder-Kieffer C.A., Buonaccorsi J., Development and validation of an estuarine biotic integrity index, Estuaries 20 (1997) 601–617. [16] Gelwick F.G., Longitudinal and Temporal Comparisons of Riffle and Pool Fish Assemblages in a Northeastern Oklahoma Ozark Stream, Copeia 4 (1990) 1072–1082. [17] Bianchi G., Demersal assemblages of the continental shelf and slope edge between the Gulf of Tehunantepec (Mexico) and the Gulf of Papagayo (Costa Rica), Mar. Ecol.-Prog. Ser. 73 (1991) 121–140. [18] Laedsgaard P., Johnson C.R., Mangrove habitats as nurseries: unique assemblages of juvenile fish in subtropical mangroves in eastern Australia, Mar. Ecol.-Prog. Ser. 126 (1995) 67–81.
Acknowledgements. We would like to thank M. Louis, C. Bouchon and Y. Bouchon-Navarro in their precious help during the collection of data, and D. Chessel and M. Simier for their constructive criticisms in the data analyses.
[27] Rooker J.R., Vulnerability of newly settled red drum (Sciaenops ocellatus) to predatory fish: Is early-life survival enhanced by seagrass meadows? Mar. Biol. 131 (1998) 145–151. [28] Vargas-Maldonado I., Yanez-Arancibia A., Fish community structure in sea grasses system (Thalassia testudinum) from Terminos Lagoon, Campeche, Mexico. Anal. Inst. Cienc. Mar Limnol. Univ. Nacion. Auton. Mexico 14 (1987) 181–196. [29] Bellwood D.R., Ontogenetic changes in the diet to early postsettlement Scarus species (Pisces: Scaridae), J. Fish Biol. 33 (1988) 213–219. [30] Thayer G.W., Bjorndal K.A., Ogden J.C., Williams S.L., Zieman J.C., Role of larger herbivores in seagrass communities, Estuaries 7 (1984) 351–376. [31] Targett T.E., Targett N.M., Energetics of food selection by the herbivorous parrotfish Sparisoma radians: Role of assimilation efficiency, gut evacuation rate, and algal secondary metabolites, Mar. Ecol.-Prog. Ser. 66 (1990) 13–22. [32] Louis M., Guyard A., Contribution à l’étude des peuplements ichtyologiques dans les mangroves de Guadeloupe (Antilles françaises), B. Ecol. 1 (1982) 9–24. [33] Louis M., Lasserre G., Etude de peuplement de poissons dans les lagunes des mangroves de Guadeloupe (Antilles Françaises), Oceanol. Acta n ° spécial sept. (1982) 333–338. [34] Louis M., Baelde P., Fish communities in Guadeloupe mangroves, Proc. V Congr. Europ. Ichthyol, Stockholm, 1985, pp. 175–180. [35] Louis M., Lam Hoai T., Lasserre G., Résultats préliminaires sur le recrutement en poissons dans deux lagunes des mangroves de Guadeloupe: Belle-Plaine et Manche-à-l’eau, Rev. Hydrobiol. Trop. 18 (1985) 249–265. [36] Galzin R., Les extractions de sable corallien en Guadeloupe et leurs conséquences sur la faune ichtyologique, C.R. Acad. Sci., ser. III 289 (1979) 853–856.
645
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646 [37] Galzin R., Toffart J.L., Louis M., Guyard A., Contribution à la connaissance de la faune ichtyologique du Grand Cul-de-Sac Marin en Guadeloupe, Cybium 6 (1982) 85–89. [38] Bouchon C., Bouchon-Navaro Y., Imbert D., Louis M., The effects of hurricane Hugo on the coastal environment of Guadeloupe Island French West Indies, Ann. Inst. Oceanogr. 67 (1992) 5–34. [39] Aliaume C., Lasserre G., Louis M., Organisation spatiale du peuplement ichtyologique des herbiers à Thalassia du Grand Cul-de-Sac Marin en Guadeloupe, Rev. Hydrobiol. Trop. 23 (1990) 231–250. [40] Aliaume C., Monteiro C., Louis M., Lam Hoai T., Lasserre G., Organisation spatio-temporelle des peuplements ichtyologiques dans deux lagunes côtières: au Portugal et en Guadeloupe, Oceanol. Acta 16 (1993a) 291–301. [41] Aliaume C., Louis M., Lasserre G., L’écosystème des herbiers à Thalassia testudinum en Guadeloupe (Antilles françaises): Relations entre le recrutement des poissons et l’habitat, Acta Oecol. 14 (1993b) 627–641. [42] Stoner A.W., Community structure of the demersal fish species of Laguna Joyuda, Puerto Rico, Estuaries 9 (1986) 142–152. [43] Field J.G., Clarke K.R., Warwick R.M., A practical strategy for analysing multispecies distribution patterns, Mar. Ecol.-Prog. Ser. 8 (1982) 37–52. [44] Dolédec S., Chessel D., Rythmes saisonniers et composantes stationnelles en milieu aquatique I– Description d’un plan d’observations complet par projection de variables, Acta Oecol. 8 (1987) 403–426. [45] Dolédec S., Chessel D., Rythmes saisonniers et composantes stationnelles en milieu aquatique II– Prise en compte et élimination d’effets dans un tableau faunistique, Acta Oecol. 10 (1989) 207–232. [46] Manly B.F.J., Randomization and Monte Carlo methods in biology, Chapman and Hall, London, 1991. [47] Dolédec S., Chessel D., Co-inertia analysis: an alternative method for studying species–environment relationships, Freshwater Biol. 31 (1994) 277–294. [48] Persat H., Olivier J.M., Pont D., Theoretical habitat templets, species traits, and species richness: Fish in the Upper Rhone River and its floodplain, Freshwater Biol. 31 (1994) 439–453. [49] Franquet E., Dolédec S., Chessel D, Using multivariate analyses for separating spatial and temporal effects within species–environment relationships, Hydrobiologia 300–301 (1995) 425–431. [50] Castella E., Speight M.D.C., Knowledge representation using fuzzy coded variables: An example based on the use of Syrphidae (Insecta, Diptera) in the assessment of riverine wetlands, Ecol. Model. 85 (1996) 13–25. [51] Bournaud M., Cellot B., Richoux P., Berrahou A., Macroinvertebrate community structure and environmental characteristics along a large river: Congruity of patterns for identification to species or family, J. North Am. Benthol. Soc. 15 (1996) 232–253. [52] Towsend C.R., Scarsbrook M.R., Dolédec S., Quantifying disturbance in streams: Alternative measures of disturbance in relation to macroinvertebrate species traits and species richness, J. North Am. Benthol. Soc. 16 (1997) 531–544.
646
[53] Statzner B., Hoppenhaus K., Arens M.F., Richoux P., Reproductive traits, habitat use and templet theory: a synthesis of world-wide data on aquatic insects, Freshwater Biol. 37 (1997) 463–472. [54] Bournaud M., Tachet H., Berly A., Cellot B., Importance of microhabitat characteristics in the macrobenthos microdistribution of a large river reach, Ann. Limnol. 34 (1998) 83–98. [55] Merigoux S., Ponton D., Spatio-temporal distribution of young fish in tributaries of natural and flow regulated sections of a neotropical river in French Guiana, South America, Freshwater Biol. 41 (1999) 1–22. [56] Franquet E., Chessel D., Approche statistique des composantes spatiales et temporelles de la relation faune-milieu, C.R. Acad. Sci. Ser. III 317 (1994) 202–206. [57] Thioulouse J., Chessel D., Dolédec S., Olivier J.M., ADE-4: a multivariate analysis and graphical display software, Statistics and Computing 7 (1997) 75–83. [58] Randall J.E., Caribbean reef fishes, T.F.H. Publications, INC, 2nd edition, 1983. [59] Connell J.H., Diversity in tropical rain forests and coral reefs, Science 199 (1978) 1302–1310. [60] Choat J.H., Ayling A.M., Schiel D.R., Temporal and spatial variation in an island fish fauna, J. Exp. Mar. Biol. Ecol. 121 (1988) 91–111. [61] Fowler A.J., Spatial and temporal patterns of distribution and abundance of chaetodontid fishes at One Tree reef, southern GBR, Mar. Ecol.-Prog. Ser. 64 (1990) 39–53. [62] Green A.L., Spatial, temporal and ontogenic patterns of habitat use by coral reef fishes (family Labridae), Mar. Ecol.-Prog. Ser. 133 (1996) 1–11. [63] Augustin D., Galzin R., Legendre P., Salvat B., Variation interannuelle des peuplements récifaux du récif-barrière de Tiahura (île de Moorea, Polynésie française), Oceanol. Acta 20 (1997) 743–756. [64] Holbrook S.J., Kingsford M.J., Schmitt R.J., Stephens J.S. Jr., Spatial and temporal patterns in assemblages of temperate reef fish, American Zoologist 34 (1994) 463–475. [65] Yanez-Arancibia A., Lara-Domiguez A.L., Sanchez-Gill P., Vargas Maldonado P., Chavance F., Amezcua linares F., Aguirre leon A., Diaz Ruiz S., Ecosystem dynamics and nichthemeral and seasonal programming of fish community structure in a tropical estuarine inlet, Mexico, Oceanol. Acta, n ° spécial, sept. (1982) 417–429. [66] Sale P.F., Guy J.A., Steel W.J., Ecological stucture of assemblages of coral reef fishes on isolated patch reefs, Oecologia 98 (1994) 83–99. [67] Letourneur Y., Chabanet P., Variations spatio-temporelles de l’ichtyofaune de platiers récifaux à la Réunion, Cybium 18 (1994) 25–38. [68] Day J.W. Jr., Hall C.A.S., Kemp M.W., Yanez-Arancibia A., Estuarine ecology, John Willey & Sons, New-York, Chichester, Brisbane, Toronto, Singapore, 1989. [69] Letourneur Y., Kulbicki M., Labrosse P., Spatial structure of commercial reef fish communities along a terrestrial runoff gradient in the northern lagoon of New Caledonia, Environ. Biol. Fish. 51 (1998) 141–159.
Ecology / Écologie
Spatial and temporal co-structure analyses between ichthyofauna and environment: an example in the tropics Laurence Blanc, Catherine Aliaume*, Alfonso Zerbi, Gérard Lasserre Université Montpellier II, laboratoire hydrobiologie marine et continentale, CNRS UMR 5556, CC 093, place E.-Bataillon, 34095 Montpellier cedex 05, France Received 20 October 2000; accepted 12 March 2001 Communicated by Claude Combes
Abstract – Ichthyofauna distribution and habitat characteristics of Thalassia beds in the Grand Cul-de-Sac marin lagoon in Guadeloupe were studied during a one-year survey. Environmental variables (9) were measured monthly in ten sites along with collection of fish communities. The environmental data set, analysed alone through between–within group ‘principal component analysis’ (PCA), exhibited a significant spatial and temporal variability. The fish data set, however, presented only a significant spatial structure, stable over the year. Given the lack of temporal variability in fish distribution, a ‘between-site co-structure analysis’ (BSCA) was used to compare the faunistic and environmental structures in space. The co-inertia structure was reduced to one axis representing a strong coast-reef gradient, the major common phenomena to both data sets. Environment and fish distribution allowed to distinguish sites directly under mangrove influence (characterised by high seagrasses, high concentration of chlorophyll a and high densities of zooplankton), to sites under reef influence (with short but dense seagrasses, clear water, and poor nutriments). For that purposes, the BSCA summarised efficiently what in common the fauna spatial structure and the environment spatial structure may present. © 2001 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS fish community / species-environment relationships / co-inertia analysis / seagrass / Guadeloupe
Résumé – Analyse de la co-structure spatiale et temporelle ichtyofaune–environnement : un exemple en milieu tropical. La distribution de l’ichtyofaune et les caractéristiques environnementales de dix sites d’herbiers à Thalassia du Grand Cul-de-Sac Marin en Guadeloupe ont été étudiées mensuellement pendant 12 mois. Le fichier des variables environnementales, traité séparément par analyse en composantes principales (ACP) inter–intragroupes, a présenté une variabilité spatiale et temporelle significative. En revanche, le fichier des biomasses par espèces de poissons a présenté seulement une variabilité spatiale significative, stable dans le temps. Étant donné le manque de variabilité temporelle pour les poissons, seule une analyse de co-structure inter-station (ACIS) a été tentée entre les deux jeux de données. Les résultats de l’analyse de co-inertie spatiale ont permis de mettre en évidence un axe fort, commun aux deux tableaux, qui représente un gradient côte–récif, avec des sites sous l’influence de la mangrove colonisés d’herbiers de grande taille, de fortes concentration de chlorophylle a et de zooplancton, opposés aux sites davantage sous l’influence de récifs coralliens
*Correspondence and reprints. E-mail address: [email protected] (C. Aliaume).
635
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
avec des herbiers plus courts mais denses, une eau plus claire mais pauvre en nutriments. Dans le cas présenté, l’analyse de co-inertie synthétise efficacement ce que la structure spatiale des communautés de poissons peut avoir en commun avec la structure spatiale de l’environnement. © 2001 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS communauté de poissons / relations espèces–environnement / analyse de co-inertie / herbiers / Guadeloupe
Version abrégée
Dans le but d’étudier l’ichtyofaune des herbiers à Thalassia du Grand cul-de-sac Marin (Guadeloupe) et ses relations avec les descripteurs de l’environnement, dix stations ont été échantillonnées mensuellement de décembre 1987 à novembre 1988, simultanément aux mesures de sept descripteurs de l’environnement : densité de plants de Thalassia, longueur moyenne de feuille, salinité, transparence, température de l’eau, concentration en nitrates, silicates, chlorophylle a, et densité en copépodes. Dans un premier temps, l’analyse spatiale et/ou temporelle des données a été réalisée par ANOVA à deux facteurs, puis complétée par ACP inter- et intraclasses (dates ou stations). Les pourcentages de variabilité inter-classes obtenus sont ensuite jugés significatifs ou non par des tests de permutations. De cette étude préalable des variables, on conclut que le milieu n’est ni stable dans le temps, ni homogène dans l’espace. Transparence, température, salinité et chlorophylle a varient significativement (p < 0,01) dans l’espace et dans le temps. Densité en copépodes, longueur et densité des herbiers présentent une variabilité spatiale significative (p < 0,01). Seuls les nitrates et les phosphates ne présentent pas de variabilité significative ni dans l’espace ni dans le temps. L’ACP inter-dates des données environnementales oppose le long de son premier axe factoriel (68 %) les mois de la saison sèche aux mois de la saison humide, avec la température et la salinité pour descripteurs saisonniers. L’analyse inter-stations permet d’opposer le long du premier axe (65 %) les stations du large aux stations côtières grâce essentiellement aux descripteurs de struc-
1. Introduction Coastal aquatic ecosystems (such as lagoons, estuaries, creeks or marsh) are inhabited by marine, fresh-water or brackish fish populations which exhibit various habitat use strategies: they may be migrant or resident, using the habitat as juvenile or adult, for reproduction or foraging ground. Most of these ecosystems play an important role as nurseries for numerous species of fish, offering food and shelter to the larval and juvenile stages [1–6]. It is also well documented that fish species are distributed depending on
636
ture des herbiers et aux concentrations de chlorophylle a et copépodes. Sur le second axe (19 %) apparaît essentiellement un gradient de salinité. L’ANOVA à deux facteurs réalisée sur les espèces de poissons a conclu à un effet mois significatif dans 15 % des cas et à un effet station significatif dans plus de 80 % des cas. La faune n’est donc pas homogène dans l’espace, elle est en revanche plutôt stable dans le temps. Lors de l’ACP inter-date, la structure spatiale faunistique a fait ressortir un gradient côte–récif très net. Dans les stations plus au large, des espèces récifales sont mélangées à des espèces plus typiques d’herbiers. Dans les stations côtières, on a pu distinguer un peuplement sous l’influence plus marquée de la mangrove, d’un peuplement à double tendance récifale et mangrove que l’on peut qualifier d’intermédiaire. Cohérente avec les analyses séparées, l’analyse de co-structure spatiale a mis en évidence un gradient côte–récif assez marqué, permettant de distinguer les stations directement sous l’influence de la mangrove caractérisées par des herbiers longs et de fortes concentrations en chlorophylle a et copépodes, des stations sous influence récifale caractérisées par une forte transparence des eaux et des herbiers denses. Le gradient côte–récif représente le principal phénomène commun aux deux tableaux de données (un seul axe représentant 93 % de l’inertie initiale). Le gradient de salinité qui jouait un rôle dans la structure spatiale du milieu, n’apparaît plus dans la co-structure spatiale révélant ainsi que la distribution spatiale des espèces de poissons n’est pas liée aux variations spatiales de ce facteur.
environmental gradients, life cycles and species interactions. Among the environmental descriptors showing a significant effect in fish community structure of coastal systems we may cite: temperature [7–10], salinity [8, 9, 11–13], dissolved oxygen [14, 15], water depth [16–18], turbidity [2, 19, 20], morphological characteristics of the substrate [21–23]. Our understanding of the relationships between habitat and fauna distribution is a corner stone in determining habitat value and productivity potential. The present study deals with the fish community associated with seagrass ecosystem of the Grand Cul-de-Sac
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
marin, the largest lagoon of Guadeloupe (French West Indies). Seagrasses, largely composed of Thalassia testudinum Banks ex Konig, are known to serve as nursery ground for numerous fish species [24–27]. It serves also as a passage way for migrating species [28] and may be used as direct food source by herbivorous species [29–31]. If fish in mangrove [32–35] and in coral reefs [36–38] have been studied for decades in Guadeloupe, it is only recently that scientific interest has focused on fish communities inhabiting seagrass beds [25, 39–41]. In this study, we characterize fish assemblages associated with the Thalassia beds in space and time and we hypothesize that there should be some agreement between the ichthyofauna structure and the environment structure.
As a consequence the objectives of this study are 1) to determine spatial and temporal structure of fauna and habitat data sets in a separate analysis, and 2) to demonstrate the agreement between faunistic and environmental data.
2. Materials and methods 2.1. Data and site description
The Grand Cul-de-Sac marin lagoon of Guadeloupe island is a large bay (11 000 ha) limited seaward by a 30-km-long barrier reef and landward by an important mangrove forest (figure 1). Seagrasses have settled in this
Figure 1. Sampling location of the Grand-Cul-de-Sac-Marin (Guadeloupe, FWI)
637
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
differences in space and time of the environmental and biological factors. To comply with normality assumption, simple normalising transformations (Log (x), X0,5) were performed on environmental variables.
well protected area and usually occupy habitats from very shallow water (less than 0.5 m deep) to 3 m of depth. During a stratified pre-sampling survey [41], seagrasses (Thalassia) were studied (length, densities and biomass recorded) on 20 sites and analysed to determine stations with redundant characteristics. Some sites had to be eliminated to comply with fishing gear constraints (depth limited to 1.5 m, no presence of corals). As a result, ten sites, where seagrass characteristics showed significant differences, were selected (figure 1): three sites were adjacent to a coral reef barrier (I, J, M), four sites were near a mangrove fringe (N, P, Q, R), one site was near a river (C) and two sites were inside the lagoon (A, B). All sites were sampled monthly from December 1987 to November 1988. Fish fauna was collected using a bag seine (50 m long, 2 m high with a square mesh ranging from 10 mm at the aisle to 3 mm in the bag). Three hauls were pulled at each site since this effort was demonstrated to be the best compromise between sampling cost and fauna representativity [39]. To eliminate the probable lunar influence, samples were always collected in the morning during the last quarter phase, which corresponded to the maximum sampling vulnerability [42]. The final fauna data set showed species biomass collected in 120 samples. One hundred species were identified, however, only species representing at least 0.1 % of the total biomass and present at least in 5 % of samples (38 species) were kept in the present analyses to avoid bias associated with rare species. Environmental factors chosen for their potential direct or indirect effect on fish distribution were: Thalassia blade density (number of blades per m2 – DENS); blade length (cm – LENG); transparency, coded from 1 to 12 with a Secchi disk (TRAN); surface temperature (taken 30 cm below surface – °C – TEMP); surface salinity (taken 30 cm below surface – SAL); nitrate concentration (mg·L–1 – NO3); silicate concentration (mg·L–1 – SiO2); chlorophyll a (mg·m–3 – CLA); copepod density (numbers·m–3 – COP), representing 90 % of the zooplankton density. These descriptors were measured simultaneously to the fish collection at all sites, except for the seagrass density and length, which were measured only at the beginning of the survey and supposed to remain stable over the year of the study.
Multivariate analyses such as between-group / withingroup ‘principal component analyses’ [44, 45] use time and space as instrumental variables that control the analysis. Between-group analysis focuses on the difference between groups (e.g. between sites); it seeks for axes that will discriminate best the centres of gravity of each group. For example, in our study, the between-site analysis of the fauna (X1, 38 species, 120 samples) will be a PCA of the site mean set (X1m, 38 species, 10 site means). In order to highlight the variability of each sample around the group centre, the initial data of table I are used as supplementary individuals. Within-group analysis seeks for axes shared by the groups. Therefore the within-site analysis of the fauna will be a PCA of the data set centred by site (X1–, 38 species, 120 centred by site values). The statistical significance of between-group analysis is tested using a MonteCarlo permutation test [46]. To answer the question of whether there is an agreement between the structure defined by the fauna and the structure defined by the environment one can use the co-inertia analysis. This analysis enables to examine species–environment relationships when many species and environmental variables are sampled in a small number of sites. The agreement between the two data sets is evaluated by determining the co-structure. The method calculates factorial axes by maximising the covariance between the factorial scores of the two data sets [47]. So, the analysis maximises the variance, which defines the structure of each table separately but also, the correlation between the two new sets of projected co-ordinates. The advantage of co-inertia analysis is demonstrated by its extensive use in freshwater biology [48–55]. Moreover between-site analysis and co-inertia analysis may be combined to study the spatial co-structure between species and their environment [56, 49]. This between-site co-structure analysis (BSCA) consists then in finding out a combination in each table of average values maximising the covariance among the between-site environmental axes and the between-site faunistic axes. To do so we first reduced the two data sets to their site mean sets, and then calculated the vectorial covariance between environmental and ichthyological variables. Finally we performed a PCA on this new matrix to define new factorial axes.
2.2. Data processing
Prior to the data processing fish biomass was transformed using double square-root to prevent abundant species from dominating the analysis as advised by Field et al. [43]. A preliminary two-way ANOVA was used to test
Table I. Environmental characteristics. N = 120 (12 months × 10 sites). TRAN = transparency; TEMP = temperature; SAL = salinity; NO3 = nitrate; SIO2 = silicate; CLA = chlorophyll a; COP = copepod density; DENS = number of blades m–2; LENG = seagrass length.
Factor mean standard-deviation
638
TRAN
TEMP (°C)
SAL
NO3 (mg·L–1)
SIO2 (mg·L–1)
CLA (mg·m–3)
COP (m–3)
DENS (m–2)
LENG (cm)
6.7 2.7
29.1 1.5
35.4 1.9
1.6 0.3
2.2 1.5
0.9 0.8
50.3 47.4
454 150.6
27.7 7.5
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
Data treatments were performed using ADE-4 [57], a free software package available on the internet (http://pbil.univ-lyon1.fr/ADE-4/).
3. Results 3.1. Spatial and temporal analyses of the environmental data set
Means and standard deviations of the environmental factors, all sites and months included, are displayed in table I. Temporal variation of the 9 environmental factors (normalised and, for some, transformed data) are plotted by site (figure 2). The length and the density of Thalassia are constant through time since the measure was not repeated and show only spatial variability. Temperature and salinity show a seasonal pattern with lower temperature and higher salinity during the dry season (December to May), and inverse tendencies during the rainy season (June to November). An important drop of the salinity is observed in site C, due to the influence on the nearby river (Goyave River, the largest river of Guadeloupe). Transparency shows a marked spatial pattern. For the four other factors no clear pattern can be seen. However, one may notice a strong increase of SiO2 concentrations in 3 sites (C, M, N) during May and a strong increase in chlorophyll a in all sites (but specially in B, N, P, Q et R) during February. The two-way ANOVA results are given in table II. There is no significant difference between sites or between months for NO3 and SiO2, which will lead us to eliminate these two variables in the rest of the analyses. There is a significant spatial difference between the copepod density, the length and the density of Thalassia. All the other environmental variables show a significant difference in both time and space. These preliminary results indicate that the environment is not stable in time and not spatially homogeneous. Between–within-group analyses were performed for space and time effect on the environmental data set (table III). The total inertia may be split into between-date inertia (representing 25 %) and within-date inertia (75 %).
Figure 2. Monthly variation of normalised environmental variables, each curve represents one site. TRAN = Transparency; TEMP = Temperature ( °C); SAL = Salinity; NO3 = Nitrate (mg·L–1); SIO2 = Silicate (mg·L–1); CLA = Chlorophyll a (mg·m–3); COP = Copepod density (m–3); DENS = number of blades m–2; LENG = Seagrass length (cm).
Table II. Two-way ANOVA performed on transformed data of environmental variables. TRAN = transparency; TEMP = temperature; SAL = salinity; NO3 = nitrate; SIO2 = silicate; CLA = chlorophyll a; COP = copepod density; DENS = number of blades m–2; LENG = seagrass length; ns: non-significant at p < 0.01; n = 120 (12 months × 10 sites); df between sites = 9; df between month = 11.
Factor Month effect Site effect
TRAN ns p < 10–4
TEMP p < 10 ns
–4
SAL p < 10 p = 3.10–2 –4
NO3 ns ns
SIO2 ns ns
CLA –4
p = 3.10 p < 10–8
COP
DENS
LENG
ns p < 10–7
p < 10–4
p < 10–4
Table III. Inertia and first eigen value (λ1) associated with global, between- and within-group PCAs of the environment data (np : not performed due to time constant factors).
PCA
global
within-date
between-date
within-site
between-site
Inertia λ1
7 2.62
5.2 (75 %) 2.53
1.8 (25 %) 1.21
3.4 (49 %) np
3.6 (51 %) 2.29
639
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
It may also be split into between-site inertia (51 %) and within-site inertia (49 %). The within-site PCA was not performed due to constant values of environmental factors (blade lengths and densities). Permutation tests for both between-date and between-site structure are highly significant (P < 0.0001), therefore both analyses are justified. In the between-date analysis (figure 3), the first factorial axis, representing 68 % of the inertia (figure 3A), opposes dry season months (December to May) to wet season months (June to November) (figure 3C). The two environmental factors exhibiting a clear seasonal pattern are salinity and temperature which are highly correlated with the first factorial axis (figure 3B). A much lesser phenomena is expressed by the second axis (18 %) which mostly opposes February to March, chlorophyll a to copepod. The between-site analysis (figure 4) present similar explanation capacity (65 % and 19 % for the first two axes
Figure 4. Between-site PCA of the environmental data set. A: Eigen values. B: 1–2 factorial map of the variables. C: Projection of the group means (circles) linked to their corresponding samples projected as supplementary data (small squares). TRAN = Transparency; TEMP = Temperature (°C); SAL = Salinity; NO3 = Nitrate (mg·L–1); SIO2 = Silicate (mg·L–1); CLA = Chlorophyll a (mg·m–3); COP = Copepod density (m–3); DENS = number of blades m–2; LENG = Seagrass length (cm).
Figure 3. Between-date PCA of the environmental data set. A: Eigen values. B: 1–2 factorial map of the variables. C: Projection of the group means (circles) linked to their corresponding samples projected as supplementary data (small squares). Months are identified by # 1 for January… TRAN = Transparency; TEMP = Temperature (°C); SAL = Salinity; NO3 = Nitrate (mg L–1); SIO2 = Silicate (mg·L–1); CLA = Chlorophyll a (mg·m–3); COP = Copepod density (m–3); DENS = number of blades m–2; LENG = Seagrass length (cm).
640
respectively – figure 4A). Sites are ranged along the first axis following more or less a coast–reef gradient (figure 4C). Sites A, N, P, and R show similar environmental characteristics, with long blades and rather low density of seagrasses, high concentrations of chlorophyll a and copepods, and low transparency. Sites I and M have very similar habitat trends, with high transparency, short but dense seagrasses, low concentrations of chlorophyll a and copepods. Sites B, C, and Q show intermediate tendencies. According to its environmental characteristics, site J is more associated with coastal sites than seaward sites, and it is primarily due to its medium size seagrass and medium range transparency. Along the second axis, site C is opposed to J due to salinity gradient and seagrass densities.
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
3.2. Spatial and temporal analyses of the ichthyofauna data set
Compared to the environment, fish community does not show a significant temporal variation.
The 38 species used in the analyses are listed in table IV, along with their relative percentage in biomass contribution, their occurrence pattern and their preferred habitat as adult [58]. The results of the two-way ANOVA performed on the 38 species separately showed that differences between sites were significant (p < 0.01) for all species except seven (Hemiramphus brasiliensis, Caranx latus, Lutjanus griseus, Bairdiella ronchus, Scorpaena grandicornis, Sparisoma chrysopterum, and Acanthurus bahianus). However, only six species (Haemulon plumieri, Haemulon sciurus, Pseudupeneus maculatus, Eucinostomus gula, Chaetodon capistratus and Sphaeroides spengleri) show a significant temporal difference (p < 0.01). The ichthyofauna will be therefore considered stable in time but not in space.
Like for the environmental data between–within-group analyses were performed on the ichthyofauna data (table V). The global PCA and the within-date analyses are very similar since there is no temporal effect. The betweendate analysis is therefore not necessary. However, the between-site analysis, representing 38.5 % of the total inertia, is highly significant (Monte Carlo permutation test – P < 0.001). The first and the second factorial axes represent respectively, 58 % and 14 % of the between-site inertia (figure 5A). The factorial map of site projection (figure 5C) shows similar trends to the one issued from the environmental analysis particularly the proximity of sites I-M, A-N-P and B-Q-R, and the coast-reef gradient given by the first axis. Along axis 2, site J is opposed to its two
Table IV. List of fish species and their codes for graphs. [1] Relative percent of biomass contribution, [2] Occurrence status in the area (R = collected in less than 50 % of the months; C = collected in 50 to 80 % of the months; P = collected in more than 80 % of the months), [3] Preferred habitat as adult (R = Coral reefs; S = Seagrass beds; M = Mangrove; P = Pelagic), [4] Results of the 2-way ANOVA : month effect (ns = not significant p = 0.01; * = significant p < 0.01; df = 11), [5] Results of the 2-way ANOVA : site effect (ns = not significant p = 0.01; * = significant p < 0.01; df = 9),
Code
species
Hja Aly Hbr Ast Sba Hpu Sfl Cla Lan Lap Lgr Och Hbo Hch Hfl Hpl Hsc Arh Cbo Bro Pma Ear Egu Sco Cca Ele Lma Scr Sch Sra Svi Aba Ach Mci Sgr Ssp Ste Dho
Harengula jaguana Anchoa lyolepis Hemiramphus brasiliensis Atherinomorus stipes Sphyraena barracuda Hypoplectrus puella Serranus flaviventris Caranx latus Lutjanus analis Lutjanus apodus Lutjanus griseus Ocyurus chrysurus Haemulon bonariense Haemulon chrysargyreum Haemulon flavolineatum Haemulon plumieri Haemulon sciurus Archosargus rhomboidalis Calamus bojanado Bairdiella ronchus Pseudupeneus maculatus Eucinostomus argenteus Eucinostomus gula Scorpaena grandicornis Chaetodon capistratus Eupomacentrus leucostictus Lachnolaimus maximus Scarus croicensis Sparisoma chrysopterum Sparisoma radians Sparisoma viride Acanthurus bahianus Acanthurus chirurgus Monacanthus ciliatus Sphaeroides greeleyi Sphaeroides spengleri Sphaeroides testudineus Diodon holacanthus
[1]
[2]
[3]
[4]
[5]
0.6 4.4 2.1 1.0 4.9 0.9 0.9 0.2 0.2 1.0 0.2 32.4 0.6 0.3 1.1 3.1 0.6 2.2 0.3 1.0 0.4 0.3 16.4 0.1 1.5 0.7 0.1 1.7 0.7 0.2 0.1 0.3 0.2 0.5 0.1 0.9 0.8 14.8
P P C P P P P C C P P P P P P P P P C R C P P R P P C P P P P P P P P P P P
P M M/P S/R R/P/S S/R S R/P R R/M M/R S/R R/S/M R R R R M/S R M R M M S R R R R R/S S R R R S M S M M/R
ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns * * ns ns ns * ns * ns * ns ns ns ns ns ns ns ns ns ns * ns ns
* * ns * * * * ns * * ns * * * * * * * * ns * * * ns * * * * ns * * ns * * * * * *
641
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
Table V. Inertia and first eigen value (λ1) associated with global. Between- and within-group PCAs of the faunistic data.
PCA
global
within-date
between-date
within-site
between-site
Inertia λ1
41 10.51
37.3 (91 %) 10.21
3.7 (9 %) ns test
25.2 (61 %) 3.29
15.8 (39 %) 9.13
neighbouring sites I-M, and sites P-A are opposed to sites B-Q-R, which seems to reflect differences between exposed (I, M, B, Q, R) to protected (J, P, A) sites. Figure 5B (projection of the species scores) allows to characterise
site groups according to fish species composition (only species contributing to 1 % of the first axis and 1 % of the second axis are mentioned here): – sites I, J, M, are characterised by reef species (Sparisoma croicensis, Eupomacentrus leucostictus, Sparisoma radians, Pseudupeneus maculatus) mixed with seagrass species (Monacanthus ciliatus, Sparisoma viride) – sites A, N, P which are under direct influence of mangrove, present mangrove species (Eucinostomus gula, Sphaeroides testudineus, Sphaeroides greeleyi, Eucinostomus argenteus, Diodon holocanthus, Archosargus rhomboidalis, Anchoa lyolepis), mixed seagrass species (Serranus flaviventris, Hypoplectrus puella), and reef species known to spend juvenile phase in seagrasses (Haemulon bonariense, Lutjanus analis, Lachnolaimus maximus) – intermediate sites B, C, Q, R, colonised by a mix of reef species (Chaetodon capistratus, Haemulon chrysargyreum, Acanthurus bahianus), mangrove species (Lutjanus apodus, Hemiramphus brasiliensis), and widely spread species (Ocyurus chrysurus). Sphyraena barracuda and Haemulon plumieri are particular species specially represented in more protected areas like A, P and J. To help visualising species distribution, scores of gravity centres are plotted (figure 6A) along with values of biomass centred over 12 months for selected species (figure 6B). This figure shows how some species, along axis 1, colonise clearly along the shore (Eucinostomus gula) or near the barrier reefs (Scarus croicencis, Monacanthus ciliatus). 3.3. Spatial co-structure between fauna and environment
Figure 5. Between-site PCA of the faunistic data set. A: Eigen values. B: 1-2 factorial map of the species. C: Projection of the site means (circles) linked to their corresponding samples projected as supplementary data (small squares).
642
Since the fauna data set did not show a significant temporal difference, the co-inertia analysis is limited to between-site co-structure of faunistic and environmental data to study the spatial structure shared by both matrices. As expected from the separate analyses, there is a significant between-site co-structure of species and environment factors (Monte Carlo permutation test, P < 0.001). The co-structure is mainly one-dimension (F1 representing 93 % of the co-structure inertia), and score projections will be therefore on a single axis (figure 7). This high correlation (0.93) between faunistic and environmental scores on this first co-structure axis confirms the strong spatial co-structure between the two data sets. Environmental scores projected on the co-inertia axis (figure 7B) confirms the main gradient observed in the environmentalone analysis with the opposition of blade length, chlorophyll a and copepods to blade density and water transparency. Likewise, fauna scores (figure 7C) show the same trends in species distribution; with coral reef species
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
however its environmental characteristics are similar to those found in intermediate sites like B, C and Q. Site Q shows also a disagreement, its fauna is similar to the one of site R but its habitat characteristics are closer to sites B or J. The differences in site I is barely significant and more difficult to interpret. For the seven other sites the difference is not significant and they all show a good agreement between fauna and environment. It is to be noticed that extreme sites such as I, M, A, N and P exhibit less variable environment or fauna structure as compared to intermediate sites such as B, Q and C more exposed to either environment variability or fish movements.
4. Discussion
Figure 6. Ichthyofauna mapping based on the between-site PCA results. A: F1 (left) and F2 (right) score mapping of centres of gravity. B: Score mapping of biomass centred over 12 months for selected species. The surface of the circle (positive value) and of the square (negative value) is proportional to the absolute value. See for the meaning of species codes.
opposed to coastal species. The distance between the centres of gravity (white circles for faunistic data and black circles for environmental data – figure 7D) gives an estimate of the agreement between fauna and environment. A shorter distance represents a better agreement. For sites J (P < 0.0001), Q (P < 0.001), and J (P < 0.05), the difference in score means between fauna and environment is significant (Wilcoxon rank test, n = 12). Site J presents a fauna composition close to those found in sites I and M
The separate between-group analyses allow independent analyses on spatial and temporal variation of a data set (either fauna or environment). In our study such analyses led to define spatial structures for both fish and habitat data sets. The environment showed a prominent coast-reef gradient expressed by seagrass morphological structures (higher in calm waters but also less dense in muddy substrate of the coastal sites, short in agitated waters of seaward sites), and productivity levels (higher concentration of chlorophyll a and density of copepods towards the coast). Site J, even though in a seaward area, is not characteristic of his area since it is a sheltered site and presents relatively calm and turbid waters. Seasonal variations are clearly demonstrated for environmental variables such as temperature and salinity, which follow the usual pattern of rainfall in the Caribbean region. Other temporal changes, this time more drastic, may be observed for variables such as chlorophyll a and densities in zooplankton which follow the episodic pattern of blooms. The fauna distribution shows a clear spatial pattern with sites under reef influence (IJM) characterised by parrotfish (Sparisoma croicensis, S. radians, S. viride), goatfish (Pseudupeneus maculatus) and damselfish (Eupomacentrus leucostictus), and sites under mangrove influence (ANP) with fauna dominated by moharras (Eucinostomus gula), puffers (Sphaeroides testudineus, S. greeleyi), grunts (Haemulon bonariense), and seabream (Archosargus rhomboidalis). These species are well established in their habitat and present very little temporal variation over the year. They share their habitat with seasonal species like grunts (Haemulon plumieri, H. sciurus) and butterflyfish (Chaetodon capistratus). This temporal stability despite varying environmental conditions, suggests that the community structure in the Grand Cul-de-Sac marin lagoon is in ‘equilibrium’ state [59], regulated more by biological interactions, predation or resource-sharing than by environmental fluctuations. Baelde [25] explained fish community stability in Guadeloupe seagrass beds by the continuous migration of fish from coral reefs to seagrasses. Louis et al. [14] observed a relative temporal fish community stability in mangrove ecosystem when studying individual abundance or biomass. Other studies demonstrated fish stability based on the survey of species composition or
643
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
Figure 7. Between-site co-structure analysis of fauna-environment data sets. A: Eigen values. B: Environment score projection. C: Fauna score projection. D: Site mean score projection seen through the fauna (white circles) and through the environment (black circles) link to their corresponding samples.
distribution in coral reefs [60–63], or based on species richness in temperate reefs [64]. On the other hand, few studies demonstrated temporal variations of fish communities in tropical lagoons [65], in coral patch reefs [66] or in mangrove areas [9]. It is to be noticed that the present study did not take into account the biodiversity and the ephemeral entrance of rare species since we excluded from the analyses more than 60 species (biomass contribution < 0.1 % of the total catch). We voluntarily used the more abundant species to draw the general patterns of the fish communities in seagrass beds. The data collection was also limited to one year of observations which does not allow inter-annual comparisons. While temporal stability in fish community of tropical ecosystems may not be commonly observed, a spatial structure however has been demonstrated in many studies [14, 25, 60–62, 66, 67]. In our study, as both environmental and fish data showed a coherent spatial distribution, a
644
spatial co-inertia analysis was justified and expected to express what the two data sets had in common spatially. The overall variation explained in between-site co-inertia was reduced to one axis (representing 93 % of it), whereas the separate between-site analyses led to first axes explaining 65 % and 58 % of the initial inertia (for the fauna set and the habitat set respectively). This is explained by the fact that the major phenomena common to both data sets is the coast–reef gradient and that most of the sites show a good agreement between environmental characteristics and ichthyofauna composition. The environmental variables involved in that axis such as seagrass height and density, and nutriments present a coherent structure with species distribution and may be involved in structuring the fish community. Water turbidity and vegetation complexity are known to play an important role in predation escapement and juvenile survival [2, 19, 27]. The other phenomena showed in separate analyses by the second
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646
axis (salinity gradient for the habitat and ‘shelter’ gradient for the fauna) disappeared in the co-inertia analysis. This revealed that the spatial distribution of fish species in this study is not linked to the spatial variation of salinity. Salinity influence on the fish community has in fact been discussed and in some instances, species do not show preferences to salinity gradients and may be influenced to some other more important factors [68] such as water turbidity [69], dissolved oxygen [14] or habitat complexity [23]. In conclusion, the between-site co-inertia analysis
summarises efficiently what in common the fauna structure and the environment structure may present spatially, regardless of the temporal aspect.
References
[19] Cyrus D.P., Blaber S.J.M., The influence of turbidity on juvenile marine fishes in estuaries. Part 1. Field studies at Lake St. Lucia on the Southeastern coast of Africa, Exp. Mar. Biol. Ecol. 109 (1985) 53–70. [20] Bottom D.L., Jones K.K., Species composition, distribution, and invertebrate prey of fish assemblages in the Columbia River Estuary, Prog. Oceanogr. 25 (1990) 243–270. [21] Stoner A.W., Distribution of fishes in seagrass meadows: role of macrophyte biomass and species composition, Fish. B. 81 (1983) 847–856. [22] Jenkins G.P., Wheatley M.J., The influence of habitat structure on nearshore fish assemblages in a southern Australian embayment: Comparison of shallow seagrass, reef-algal and unvegetated sand habitat, with emphasis on their importance to recruitment, J. Exp. Mar. Biol. Ecol. 221 (1998) 147–172. [23] Garcia-Charton J.A., Perez-Rufaza A., Correlation between habitat structure and a rocky reef fish assemblages in the southwest Mediterranean, Mar. Ecol. 19 (1998) 111–128. [24] Livinsgton R.J., Trophic organization of fishes in a coastal seagrass system, Mar. Ecol.-Prog. Ser. 7 (1982) 1–12. [25] Baelde P., Differences in the structures of fish assemblages in Thalassia testudinum beds in Guadeloupe, French West Indies, and their ecological significance, Mar. Biol. 105 (1990) 163–173. [26] Rooker J.R., Holt S.A., Utilization of subtropical seagrass meadows by newly settled red drum Sciaenops ocellatus: Patterns of distribution and growth, Mar. Ecol.-Prog. Ser. 158 (1997) 139–149.
[1] Weinstein M.P., Shallow marsh habitats as primary nurseries for fishes and shellfish, Cape Fear River, North Carolina, Fish. B. 77 (1979) 339–357. [2] Blaber S.J.M., Blaber T.G., Factors affecting the distribution of juvenile estuarine and inshore fish, J. Fish Biol. 17 (1980) 143–162. [3] Weinstein M.P., Weiss S.L., Walters M.F., Multiple determinants of community structure in shallow marsh habitats Cape Fear River estuary, Noth Carolina, U.S.A, Mar. Biol. 58 (1980) 227–243. [4] Loneragan N.R., Potter I.C., Lenanton R.C.J., Caputi N., Spatial and seasonal differences in the fish fauna in the shallows of a large Australian estuary, Mar. Biol. 92 (1986) 575–586. [5] Tzeng W.N., Wang Y.T., Structure, composition and seasonal dynamics of the larval and juvenile fish community in the mangrove estuary of Tanshui River, Taiwan, Mar. Biol. 113 (1992) 481–490. [6] Yanez-Arancibia A., Lara-Domiguez A.L., Day J.W. Jr., Interactions between mangrove and seagrass habitats mediated by estuarine nekton assemblages: Coupling of primary and secondary production, Hydrobiologia 264 (1993) 1–12. [7] Morin B., Hudon C., Whoriskey F.G., Environmental influences on seasonal distribution of coastal and estuarine fish assemblages at Wemindji, eastern James Bay, Environ. Biol. Fish. 35 (1992) 219–229. [8] Rogers S.I., Millner R.S., Factors affecting the annual abundance and regional distribution of English inshore demersal fish populations: 1973 to 1995, ICES J. Mar. Sci. 53 (1996) 1094–1112. [9] Laroche J., Baran E., Rasoanandrasana N.B., Temporal patterns in a fish assemblage of a semiarid mangrove zone in Madagascar, J. Fish Biol. 51 (1997) 3–20. [10] Tzeng W.N., Wang Y.T., Chern Y.T., Species composition and distribution of fish larvae in Yenliao Bay, Northeastern Taiwan, Zool. Stud. 36 (1997) 146–158. [11] Rozas L.P., Odum W.E., Use of tidal marshes by fishes and macrofaunal crustaceans along a marsh stream-order gradient, Estuaries 10 (1987) 36–43. [12] Odum W.E., Comparative ecology of tidal freshwater and saltmarshes, Ann. Rev. Syst. 19 (1988) 147–176. [13] Thiel R., Sepulveda A., Kafemann R., Nellen W., Environmental factors as forces structuring the fish community of the Elbe Estuary, J. Fish Biol. 46 (1995) 47–69. [14] Louis M., Bouchon C., Bouchon-Navaro Y., Spatial and temporal variations of mangrove fish assemblages in Martinique (French West Indies), Hydrobiologia 295 (1995) 275–284. [15] Deegan L.A., Finn J.T., Ayvazian S.G., Ryder-Kieffer C.A., Buonaccorsi J., Development and validation of an estuarine biotic integrity index, Estuaries 20 (1997) 601–617. [16] Gelwick F.G., Longitudinal and Temporal Comparisons of Riffle and Pool Fish Assemblages in a Northeastern Oklahoma Ozark Stream, Copeia 4 (1990) 1072–1082. [17] Bianchi G., Demersal assemblages of the continental shelf and slope edge between the Gulf of Tehunantepec (Mexico) and the Gulf of Papagayo (Costa Rica), Mar. Ecol.-Prog. Ser. 73 (1991) 121–140. [18] Laedsgaard P., Johnson C.R., Mangrove habitats as nurseries: unique assemblages of juvenile fish in subtropical mangroves in eastern Australia, Mar. Ecol.-Prog. Ser. 126 (1995) 67–81.
Acknowledgements. We would like to thank M. Louis, C. Bouchon and Y. Bouchon-Navarro in their precious help during the collection of data, and D. Chessel and M. Simier for their constructive criticisms in the data analyses.
[27] Rooker J.R., Vulnerability of newly settled red drum (Sciaenops ocellatus) to predatory fish: Is early-life survival enhanced by seagrass meadows? Mar. Biol. 131 (1998) 145–151. [28] Vargas-Maldonado I., Yanez-Arancibia A., Fish community structure in sea grasses system (Thalassia testudinum) from Terminos Lagoon, Campeche, Mexico. Anal. Inst. Cienc. Mar Limnol. Univ. Nacion. Auton. Mexico 14 (1987) 181–196. [29] Bellwood D.R., Ontogenetic changes in the diet to early postsettlement Scarus species (Pisces: Scaridae), J. Fish Biol. 33 (1988) 213–219. [30] Thayer G.W., Bjorndal K.A., Ogden J.C., Williams S.L., Zieman J.C., Role of larger herbivores in seagrass communities, Estuaries 7 (1984) 351–376. [31] Targett T.E., Targett N.M., Energetics of food selection by the herbivorous parrotfish Sparisoma radians: Role of assimilation efficiency, gut evacuation rate, and algal secondary metabolites, Mar. Ecol.-Prog. Ser. 66 (1990) 13–22. [32] Louis M., Guyard A., Contribution à l’étude des peuplements ichtyologiques dans les mangroves de Guadeloupe (Antilles françaises), B. Ecol. 1 (1982) 9–24. [33] Louis M., Lasserre G., Etude de peuplement de poissons dans les lagunes des mangroves de Guadeloupe (Antilles Françaises), Oceanol. Acta n ° spécial sept. (1982) 333–338. [34] Louis M., Baelde P., Fish communities in Guadeloupe mangroves, Proc. V Congr. Europ. Ichthyol, Stockholm, 1985, pp. 175–180. [35] Louis M., Lam Hoai T., Lasserre G., Résultats préliminaires sur le recrutement en poissons dans deux lagunes des mangroves de Guadeloupe: Belle-Plaine et Manche-à-l’eau, Rev. Hydrobiol. Trop. 18 (1985) 249–265. [36] Galzin R., Les extractions de sable corallien en Guadeloupe et leurs conséquences sur la faune ichtyologique, C.R. Acad. Sci., ser. III 289 (1979) 853–856.
645
L. Blanc et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 324 (2001) 635–646 [37] Galzin R., Toffart J.L., Louis M., Guyard A., Contribution à la connaissance de la faune ichtyologique du Grand Cul-de-Sac Marin en Guadeloupe, Cybium 6 (1982) 85–89. [38] Bouchon C., Bouchon-Navaro Y., Imbert D., Louis M., The effects of hurricane Hugo on the coastal environment of Guadeloupe Island French West Indies, Ann. Inst. Oceanogr. 67 (1992) 5–34. [39] Aliaume C., Lasserre G., Louis M., Organisation spatiale du peuplement ichtyologique des herbiers à Thalassia du Grand Cul-de-Sac Marin en Guadeloupe, Rev. Hydrobiol. Trop. 23 (1990) 231–250. [40] Aliaume C., Monteiro C., Louis M., Lam Hoai T., Lasserre G., Organisation spatio-temporelle des peuplements ichtyologiques dans deux lagunes côtières: au Portugal et en Guadeloupe, Oceanol. Acta 16 (1993a) 291–301. [41] Aliaume C., Louis M., Lasserre G., L’écosystème des herbiers à Thalassia testudinum en Guadeloupe (Antilles françaises): Relations entre le recrutement des poissons et l’habitat, Acta Oecol. 14 (1993b) 627–641. [42] Stoner A.W., Community structure of the demersal fish species of Laguna Joyuda, Puerto Rico, Estuaries 9 (1986) 142–152. [43] Field J.G., Clarke K.R., Warwick R.M., A practical strategy for analysing multispecies distribution patterns, Mar. Ecol.-Prog. Ser. 8 (1982) 37–52. [44] Dolédec S., Chessel D., Rythmes saisonniers et composantes stationnelles en milieu aquatique I– Description d’un plan d’observations complet par projection de variables, Acta Oecol. 8 (1987) 403–426. [45] Dolédec S., Chessel D., Rythmes saisonniers et composantes stationnelles en milieu aquatique II– Prise en compte et élimination d’effets dans un tableau faunistique, Acta Oecol. 10 (1989) 207–232. [46] Manly B.F.J., Randomization and Monte Carlo methods in biology, Chapman and Hall, London, 1991. [47] Dolédec S., Chessel D., Co-inertia analysis: an alternative method for studying species–environment relationships, Freshwater Biol. 31 (1994) 277–294. [48] Persat H., Olivier J.M., Pont D., Theoretical habitat templets, species traits, and species richness: Fish in the Upper Rhone River and its floodplain, Freshwater Biol. 31 (1994) 439–453. [49] Franquet E., Dolédec S., Chessel D, Using multivariate analyses for separating spatial and temporal effects within species–environment relationships, Hydrobiologia 300–301 (1995) 425–431. [50] Castella E., Speight M.D.C., Knowledge representation using fuzzy coded variables: An example based on the use of Syrphidae (Insecta, Diptera) in the assessment of riverine wetlands, Ecol. Model. 85 (1996) 13–25. [51] Bournaud M., Cellot B., Richoux P., Berrahou A., Macroinvertebrate community structure and environmental characteristics along a large river: Congruity of patterns for identification to species or family, J. North Am. Benthol. Soc. 15 (1996) 232–253. [52] Towsend C.R., Scarsbrook M.R., Dolédec S., Quantifying disturbance in streams: Alternative measures of disturbance in relation to macroinvertebrate species traits and species richness, J. North Am. Benthol. Soc. 16 (1997) 531–544.
646
[53] Statzner B., Hoppenhaus K., Arens M.F., Richoux P., Reproductive traits, habitat use and templet theory: a synthesis of world-wide data on aquatic insects, Freshwater Biol. 37 (1997) 463–472. [54] Bournaud M., Tachet H., Berly A., Cellot B., Importance of microhabitat characteristics in the macrobenthos microdistribution of a large river reach, Ann. Limnol. 34 (1998) 83–98. [55] Merigoux S., Ponton D., Spatio-temporal distribution of young fish in tributaries of natural and flow regulated sections of a neotropical river in French Guiana, South America, Freshwater Biol. 41 (1999) 1–22. [56] Franquet E., Chessel D., Approche statistique des composantes spatiales et temporelles de la relation faune-milieu, C.R. Acad. Sci. Ser. III 317 (1994) 202–206. [57] Thioulouse J., Chessel D., Dolédec S., Olivier J.M., ADE-4: a multivariate analysis and graphical display software, Statistics and Computing 7 (1997) 75–83. [58] Randall J.E., Caribbean reef fishes, T.F.H. Publications, INC, 2nd edition, 1983. [59] Connell J.H., Diversity in tropical rain forests and coral reefs, Science 199 (1978) 1302–1310. [60] Choat J.H., Ayling A.M., Schiel D.R., Temporal and spatial variation in an island fish fauna, J. Exp. Mar. Biol. Ecol. 121 (1988) 91–111. [61] Fowler A.J., Spatial and temporal patterns of distribution and abundance of chaetodontid fishes at One Tree reef, southern GBR, Mar. Ecol.-Prog. Ser. 64 (1990) 39–53. [62] Green A.L., Spatial, temporal and ontogenic patterns of habitat use by coral reef fishes (family Labridae), Mar. Ecol.-Prog. Ser. 133 (1996) 1–11. [63] Augustin D., Galzin R., Legendre P., Salvat B., Variation interannuelle des peuplements récifaux du récif-barrière de Tiahura (île de Moorea, Polynésie française), Oceanol. Acta 20 (1997) 743–756. [64] Holbrook S.J., Kingsford M.J., Schmitt R.J., Stephens J.S. Jr., Spatial and temporal patterns in assemblages of temperate reef fish, American Zoologist 34 (1994) 463–475. [65] Yanez-Arancibia A., Lara-Domiguez A.L., Sanchez-Gill P., Vargas Maldonado P., Chavance F., Amezcua linares F., Aguirre leon A., Diaz Ruiz S., Ecosystem dynamics and nichthemeral and seasonal programming of fish community structure in a tropical estuarine inlet, Mexico, Oceanol. Acta, n ° spécial, sept. (1982) 417–429. [66] Sale P.F., Guy J.A., Steel W.J., Ecological stucture of assemblages of coral reef fishes on isolated patch reefs, Oecologia 98 (1994) 83–99. [67] Letourneur Y., Chabanet P., Variations spatio-temporelles de l’ichtyofaune de platiers récifaux à la Réunion, Cybium 18 (1994) 25–38. [68] Day J.W. Jr., Hall C.A.S., Kemp M.W., Yanez-Arancibia A., Estuarine ecology, John Willey & Sons, New-York, Chichester, Brisbane, Toronto, Singapore, 1989. [69] Letourneur Y., Kulbicki M., Labrosse P., Spatial structure of commercial reef fish communities along a terrestrial runoff gradient in the northern lagoon of New Caledonia, Environ. Biol. Fish. 51 (1998) 141–159.