Composition, distribution, and diversity of pelagic fishes around the Canary Islands, Eastern Central Atlantic

Abstract

Oceanic islands of volcanic origin have a narrow shelf and a steep slope that should lead to considerable spatial overlap among coastal and oceanic fauna. During six pelagic surveys in the Canarian archipelago, Eastern Central Atlantic, over 65,000 fishes belonging to 211 species were collected at depths between 8 and 1035 m. The mesopelagic families of the lanternfishes (Myctophidae) and the bristlemouths (Gonostomatidae) accounted for about 50% of all specimens. By multivariate classification and ordination methods four different assemblages associated with mesopelagic, epipelagic–oceanic or coastal habitats could be identified. Two of these assemblages were coastal, differing in the proportion of meso- and epipelagic species. These data indicate intense horizontal migrations of mesopelagic fishes (mainly Myctophidae) into the neritic realm and increased interactions between coastal and oceanic habitats. Alpha diversity indices were higher and dominance was lower in oceanic habitats compared to the coastal realm. No marked differences among oceanographically similar areas of the entire archipelago were found. Beta diversity as a measure of similarity among sites or samples revealed variabilities between areas south of Gran Canaria and Fuerteventura islands. A considerable heterogeneity in species distribution was found off SE Fuerteventura in an area with high hydrographic variability. Therefore, both topography and hydrography are important factors influencing the distribution and abundance of pelagic fishes in this oceanic archipelago.

Keywords:

• Atlantic
• Canary Islands
• diel migrations
• diversity
• mesopelagic fishes
• myctophidae

Published in collaboration with the University of Bergen and the Institute of Marine Research, Norway, and the Marine Biological Laboratory, University of Copenhagen, Denmark

Introduction

Apart from continents, shallow seamounts, or oceanic islands, the vast pelagic realm actually lacks any physical structure which could serve as a distribution barrier. Nevertheless, many of its inhabitants have well-defined regions of occurrence, sometimes being limited to relatively small areas. Conditions in the horizontal strata are spatially homogenous with patterns of seasonal and latitudinal variability (Robison 2004). Topographic and meso-scale hydrographic features proved to be the most influential for species distribution for islands of volcanic origin, like the Canarian archipelago, in the Eastern Central Atlantic. The special topographic conditions around such islands should intensify the interactions among coastal, oceanic, benthic, and benthopelagic organisms (Uiblein & Bordes 1999). This hypothesis was supported by Bordes et al. (1999), who reported the occurrence of two typical shelf-dwelling species (Scomber colias Gmelin, 1789 and Sardinella aurita Valenciennes, 1847) in oceanic waters together with mesopelagic fishes (e.g. Lampanyctus alatus Goode & Bean, 1896, Lobianchia dofleini (Zugmayer, 1911)), and Wienerroither (2003), who found numerous lanternfishes (e.g. Diaphus dumerilii (Bleeker, 1856), Hygophum hygomii (Luetken, 1892), Lobianchia dofleini) in coastal waters.

In addition, the Canarian archipelago shows a particular hydrology. The islands lie in the southward-directed Canary Current and are also under the influence of deep-water masses from the North Atlantic, the Mediterranean, and the Antarctic (Mittelstaedt 1983). Topologically induced deep water meanders (Longhurst 1998) and eddies (Arístegui et al. 1994, 1997) as well as seasonally varying upwellings (Uiblein et al. 1998), offshore water filaments, eddies, and island wakes (Barton et al. 1998) lead to a high complexity and small-scale hydrographic variability both horizontally and vertically. These influence chlorophyll values and primary production and hence the distribution of fishes.

The structure of mesopelagic micronekton assemblages has been the subject of several investigations (Hulley 1992; González & Sánchez 2002; Fock et al. 2004; Pusch et al. 2004a), revealing a close relationship to physical and trophic conditions. The decreasing water depth around seamounts reduces density and diversity of mesopelagic fish (Pusch et al. 2004b), which respond in different ways to the impingement with the demersal fauna (Porteiro & Sutton 2007). Auster et al. (1992) ascribe the aggregation behaviour of myctophid fishes with other pelagic fauna to predation risk reduction and more profitable search for prey. In a study using acoustics and a deployed camera for in-situ identification of pelagic fauna off Hawaiian Islands, Benoit-Bird & Au (2006) observed a highly dynamic use of adjacent oceanic and coastal habitats by mesopelagic fishes. They ascribed their findings to horizontal migrations of certain faunal groups including myctophid fishes, and found in an earlier study (Benoit-Bird & Au 2003) large differences in the overall distribution patterns and densities of mesopelagic animals, but did not investigate spatial distribution, assemblage structure, and diversity patterns at high taxonomic resolution. The understanding of the relationship between species richness and its spatial change is fundamental to the assessment of spatial diversity patterns (Koleff & Gaston 2002). To our knowledge few studies have been undertaken so far to thoroughly understand the shallow and deep pelagic fish assemblage structure of oceanic islands.

In the present study we further examine earlier hypotheses (see Bordes et al. 1999; Uiblein & Bordes 1999; Wienerroither 2003) that the narrow shelf and the steep slope around oceanic islands facilitate the use of the productive coastal zone by vertically migrating mesopelagic fishes and the increased abundance of the coastal fauna in the epipelagic oceanic realm. The objective of this study is to identify the species composition and distribution of pelagic fishes in the immediate surroundings of the Canary Islands in order to determine (1) overall and local diversity, and (2) the faunal change between individual stations, different habitats, and areas within the islands. The influence of sampling depth, bottom depth, and time of day as important ecological factors on species composition, distribution, and diversity are also discussed.

Material and methods

Six pelagic trawling surveys with the vessel B/E ‘La Bocaina’ were carried out within the Canarian Archipelago between 1997 and 2002 (month and year follow the acronym for each cruise): ‘La Bocaina 11/97’ (B), ‘ECOS 04/99’ (C), ‘Mesopelagic 05/99’ (D), ‘Pelagic 01/00’ (E), ‘Pelagic 11/00’ (F), and ‘Bocaina 03/02’ (G). The trawl tows were conducted horizontally during day and night, at different depths (epipelagic: 8–219 m; mesopelagic: 215–1009 m) and above different bottom depths from coastal (min. bottom depth 40 m) to oceanic (max. bottom depth 2525 m) waters. Exact data concerning duration, location, trawling and bottom depth, of all surveys and tows are given in Table I. The maps of the surveys and the single sampling stations, indicating the type (coastal, epipelagic or mesopelagic), the length, and the direction of the trawls are shown in Figures 1 and 2. Trawl tows overlapping the 200 m depth line which demarcates the epi- and mesopelagic realm were considered as epipelagic. Trawl tows overlapping the 200 m bottom depth line demarcating the coastal and oceanic realm were considered as coastal if the range of the bottom depth during the haul was less than 200 m and as epipelagic if the range was more than 200 m.

Figure 1.  Map of survey (a) ‘La Bocaina 11/97’ (B1–B18), and (b) ECOS 04/99’ (C1–C23) and ‘Mesopelagic 05/99’ (D1–D6) (legend as in Figure 1a). Station numbers refer to Table I. Insert in Figure 1a: the Canarian Archipelago, situated in the Eastern Central Atlantic.

Figure 2.  Map of survey (a) ‘Pelagic 01/00’ (E1–E11) and ‘Pelagic 11/00’ (F1–F19) (legend as in Figure 2b) and (b) ‘Bocaina 03/02’ (G1–G27). Station numbers refer to Table I.

Table I. Characteristics of trawl tow stations of all surveys. Epipelagic stations in normal, coastal stations in italics, and mesopelagic stations in bold letters. Stations B4, B7, B10, C1, C3, C11, C20, E4, F4, G9, and G11 were without result. See text for further explanations.

Survey/Station	Date	Time	Initial latitude, longitude	Final latitude, longitude	Trawl depth (range, in m)	Bottom depth (range, in m)
B1	09 Nov 97	20:35–21:35	28°07,46′ N 14°02,48′ W	28°05,96′ N 14°05,92′ W	71–78	1360–1413
B2	10 Nov 97	17:20–18:18	28°06,38′ N 13°57,83′ W	28°05,17′ N 14°00,42′ W	500–523	1067–1390
B3	10 Nov 97	20:12–21:12	28°10,04′ N 14°09,46′ W	28°09,72′ N 14°04,70′ W	34–69	617–1457
B4	10 Nov 97	11:50–12:24	28°10,23′ N 14°05,19′ W	28°10,02′ N 14°07,18′ W	22–35	49–85
B5	12 Nov 97	06:40–07:10	28°02,50′ N 14°26,28′ W	28°02,21′ N 14°24,23′ W	40–44	130–530
B6	13 Nov 97	06:25–06:52	28°51,55′ N 13°54,29′ W	28°50,16′ N 13°54,51′ W	28–43	656–860
B7	14 Nov 97	14:15–15:15	28°55,23′ N 13°34,57′ W	28°54,25′ N 13°38,19′ W	13–42	128–217
B8	15 Nov 97	10:20–11:00	28°58,67′ N 13°28,74′ W	28°56,98′ N 13°30,11′ W	53–83	101–120
B9	15 Nov 97	19:25–20:25	29°23,12′ N 13°12,23′ W	29°20,37′ N 13°11,96′ W	527–557	1280–1350
B10	16 Nov 97	00:10–01:23	29°20,42′ N 13°21,88′ W	29°16,58′ N 13°23,92′ W	20–64	94–144
B11	16 Nov 97	03:05–04:05	29°22,50′ N 13°33,36′ W	29°19,72′ N 13°33,59′ W	34–46	586–692
B12	16 Nov 97	23:25–00:35	28°50,06′ N 13°53,01′ W	28°47,27′ N 13°52,11′ W	31–43	70–115
B13	18 Nov 97	06:05–07:05	28°07,95′ N 14°01,63′ W	28°06,34′ N 14°04,23′ W	90–107	1183–1442
B14	20 Nov 97	19:45–20:45	27°38,30′ N 15°37,64′ W	27°36,98′ N 15°39,98′ W	592–603	1638–1667
B15	20 Nov 97	23:50–00:50	27°40,99′ N 15°44,21′ W	27°43,36′ N 15°46,66′ W	28–63	116–344
B16	21 Nov 97	21:20–22:20	28°07,23′ N 15°47,59′ W	28°05,07′ N 15°50,17′ W	42–66	104–112
B17	23 Nov 97	22:08–23:08	27°50,20′ N 15°19,27′ W	27°52,80′ N 15°18,27′ W	30–52	136–231
B18	24 Nov 97	00:25–01:00	27°51,58′ N 15°18,71′ W	27°49,87′ N 15°18,71′ W	116–146	181–260
C1	08 Apr 99	19:23–20:50	28°51,75′ N 13°44,02′ W	28°49,01′ N 13°46,02′ W	28–39	117–220
C2	09 Apr 99	21:17–22:00	28°50,30′ N 13°51,52′ W	28°47,30′ N 13°52,67′ W	8–39	79–99
C3	10 Apr 99	05:13–06:00	28°49,69′ N 13°52,47′ W	28°47,29′ N 13°52,89′ W	25–39	98–112
C4	10 Apr 99	21:28–23:30	28°10,07′ N 14°00,31′ W	28°09,57′ N 14°07,11′ W	19–50	102–695
C5	11 Apr 99	09:45–11:40	28°07,29′ N 14°03,65′ W	28°05,62′ N 14°09,88′ W	382–511	1264–1351
C6	12 Apr 99	22:13–23:41	28°05,76′ N 14°03,85′ W	28°04,43′ N 14°08,34′ W	23–52	1400–1544
C7	13 Apr 99	21:51–22:55	28°08,43′ N 14°12,47′ W	28°05,15′ N 14°15,71′ W	19–40	67–314
C8	14 Apr 99	05:38–06:45	28°07,85′ N 14°02,32′ W	28°06,33′ N 14°06,65′ W	47–67	1254–1359
C9	14 Apr 99	21:20–22:15	28°49,61′ N 13°45,97′ W	28°46,41′ N 13°47,25′ W	18–33	87–165
C10	15 Apr 99	10:35–12:25	28°49,82′ N 13°37,89′ W	28°45,50′ N 13°41,62′ W	258–600	1031–1077
C11	15 Apr 99	15:02–15:55	28°51,30′ N 13°41,63′ W	28°48,16′ N 13°43,95′ W	12–40	756–770
C12	22 Apr 99	22:35–23:25	27°42,78′ N 15°49,74′ W	27°39,77′ N 15°48,58′ W	38–49	1145–1315
C13	23 Apr 99	01:05–02:05	27°46,05′ N 15°48,70′ W	27°44,15′ N 15°46,49′ W	33–53	82–90
C14	23 Apr 99	16:55–18:40	27°38,99′ N 15°39,43′ W	27°36,21′ N 15°43,42′ W	480–694	1069–1384
C15	23 Apr 99	21:50–22:45	27°43,08′ N 15°43,70′ W	27°41,13′ N 15°42,58′ W	24–40	57–115
C16	24 Apr 99	00:00–01:20	27°43,78′ N 15°47,00′ W	27°41,37′ N 15°43,46′ W	18–61	100–123
C17	25 Apr 99	22:40–23:45	28°11,38′ N 16°50,51′ W	28°07,53′ N 16°47,98′ W	25–47	74–132
C18	26 Apr 99	01:01–02:09	28°07,60′ N 16°46,16′ W	28°10,95′ N 16°50,31′ W	45–71	75–176
C19	26 Apr 99	12:10–13:50	28°10,71′ N 16°53,77′ W	28°05,89′ N 16°51,04′ W	553–716	1213–1409
C20	30 Apr 99	07:06–08:00	28°04,60′ N 17°21,20′ W	28°02,37′ N 17°19,43′ W	34–52	90–107
C21	30 Apr 99	17:13–18:20	28°18,98′ N 17°06,61′ W	28°18,09′ N 17°03,89′ W	255–535	944–1283
C22	30 Apr 99	21:52–23:50	28°06,84′ N 17°23,31′ W	28°02,36′ N 17°19,96′ W	25–70	110–118
C23	01 May 99	21:15–23:10	28°41,90′ N 17°59,46′ W	28°34,61′ N 17°56,36′ W	18–42	96–784
D1	11 May 99	05:55–07:05	28°39,50′ N 13°26,49′ W	28°36,90′ N 13°29,24′ W	38–40	1327–1349
D2	11 May 99	08:35–09:40	28°34,57′ N 13°31,57′ W	28°31,85′ N 13°34,12′ W	495–561	1287–1306
D3	11 May 99	22:15–23:15	28°49,29′ N 15°22,68′ W	28°46,86′ N 15°24,52′ W	42–47	1646–1782
D4	12 May 99	09:05–10:05	28°48,30′ N 15°23,67′ W	28°46,40′ N 15°25,80′ W	517–591	1940–2525
D5	12 May 99	21:35–22:35	29°04,05′ N 16°56,10′ W	29°02,46′ N 16°58,30′ W	448–526	1660–2120
D6	12 May 99	23:45–00:45	29°02,45′ N 16°58,18′ W	29°00,60′ N 17°00,20′ W	48–66	1463–2119
E1	19 Jan 00	18:30–19:55	27°56,97′ N 15°16,52′ W	28°00,37′ N 15°18,07′ W	467–566	765–845
E2	19 Jan 00	22:10–23:10	28°00,05′ N 15°20,45′ W	28°03,48′ N 15°21,56′ W	17–50	105–294
E3	22 Jan 00	14:35–15:50	28°12,13′ N 15°31,58′ W	28°12,62′ N 15°35,50′ W	356–402	949–1455
E4	22 Jan 00	23:15–23:45	28°11,49′ N 15°38,13′ W	–	37–66	60–122
E5	25 Jan 00	16:25–18:00	27°41,51′ N 15°31,39′ W	27°40,27′ N 15°34,77′ W	490–510	950–1260
E6	25 Jan 00	20:21–21:15	27°44,58′ N 15°24,40′ W	27°43,82′ N 15°28,42′ W	10–70	76–186
E7	26 Jan 00	15:20–16:40	28°04,16′ N 15°55,10′ W	28°02,39′ N 15°56,49′ W	578–610	1239–1644
E8	26 Jan 00	20:15–21:35	28°05,24′ N 15°49,56′ W	28°02,71′ N 15°53,57′ W	40–45	101–280
E9	27 Jan 00	15:25–16:55	27°49,08′ N 15°54,01′ W	27°51,64′ N 15°56,73′ W	493–503	767–1096
E10	27 Jan 00	20:15–20:45	27°41,65′ N 15°42,91′ W	27°42,36′ N 15°44,61′ W	25–34	75–86
E11	27 Jan 00	21:45–22:45	27°41,82′ N 15°43,27′ W	27°43,23′ N 15°46,05′ W	41–64	75–92
F1	10 Nov 00	20:38–21:28	28°10,69′ N 15°31,73′ W	28°10,60′ N 15°34,48′ W	55–90	132–853
F2	12 Nov 00	18:28–19:17	28°03,04′ N 15°21,37′ W	28°00,52′ N 15°20,28′ W	17–45	103–263
F3	12 Nov 00	21:59–22:46	28°03,38′ N 15°20,70′ W	28°00,94′ N 15°20,73′ W	43–90	98–414
F4	14 Nov 00	11:00–11:52	27°59,53′ N 15°20,53′ W	27°57,86′ N 15°21,14′ W	37–64	102–240
F5	14 Nov 00	22:33–23:13	27°44,96′ N 15°26,21′ W	27°44,22′ N 15°28,23′ W	20–48	59–75
F6	15 Nov 00	21:20–22:03	27°41,04′ N 15°42,19′ W	27°41,85′ N 15°44,83′ W	37–64	98–202
F7	15 Nov 00	23:04–23:53	27°42,21′ N 15°44,79′ W	27°43,75′ N 15°46,99′ W	47–71	92–112
F8	17 Nov 00	17:21–18:39	27°38,12′ N 15°47,02′ W	27°39,52′ N 15°50,05′ W	434–606	927–1653
F9	19 Nov 00	11:00–12:14	27°37,49′ N 15°45,05′ W	27°39,54′ N 15°47,82′ W	622–813	928–1287
F10	19 Nov 00	13:28–14:38	27°40,32′ N 15°48,57′ W	27°41,92′ N 15°51,22′ W	310–530	1174–1650
F11	19 Nov 00	18:24–20:04	27°41,39′ N 15°43,84′ W	27°44,76′ N 15°48,54′ W	24–57	98–228
F12	20 Nov 00	18:28–19:35	28°02,70′ N 15°56,76′ W	28°01,24′ N 15°58,53′ W	444–900	1634–1765
F13	20 Nov 00	21:16–22:18	28°03,32′ N 15°51,38′ W	27°59,90′ N 15°53,30′ W	37–46	88–117
F14	21 Nov 00	12:08–13:23	27°37,21′ N 15°39,71′ W	27°35,82′ N 15°42,30′ W	495–1009	1344–1620
F15	21 Nov 00	15:01–16:01	27°36,26′ N 15°40,36′ W	27°37,30′ N 15°38,11′ W	592–637	1316–1688
F16	21 Nov 00	18:36–19:10	27°33,81′ N 15°21,66′ W	27°32,73′ N 15°19,18′ W	29–31	2189–2438
F17	21 Nov 00	21:30–22:34	27°37,91′ N 15°38,55′ W	27°36,44′ N 15°41,52′ W	31–37	1305–1603
F18	22 Nov 00	11:53–12:53	27°36,13′ N 15°48,77′ W	27°35,26′ N 15°46,87′ W	464–924	1440–1690
F19	22 Nov 00	14:50–15:55	27°36,01′ N 15°46,38′ W	27°37,05′ N 15°48,94′ W	496–630	1358–1564
G1	07 Mar 02	22:34–23:52	28°06,91′ N 14°13,87′ W	28°03,95′ N 14°16,75′ W	24–60	140–284
G2	08 Mar 02	19:54–20:59	28°09,49′ N 14°00,21′ W	28°10,33′ N 14°03,85′ W	13–54	217–606
G3	09 Mar 02	17:35–19:00	28°04,15′ N 14°14,83′ W	27°59,83′ N 14°17,44′ W	440–665	1107–1327
G4	09 Mar 02	21:55–23:03	28°08,22′ N 14°03,44′ W	28°07,31′ N 14°07,31′ W	22–130	1000–1211
G5	10 Mar 02	00:10–01:12	28°04,46′ N 14°14,80′ W	28°02,18′ N 14°17,21′ W	34–123	747–1006
G6	10 Mar 02	10:39–11:05	28°08,50′ N 14°02,93′ W	28°07,89′ N 14°05,88′ W	617–915	980–1162
G7	11 Mar 02	21:20–21:54	28°09,89′ N 14°01,28′ W	28°09,79′ N 14°03,55′ W	98–219	137–464
G8	12 Mar 02	20:35–21:35	28°08,54′ N 14°06,45′ W	28°07,36′ N 14°08,74′ W	269–529	656–1026
G9	13 Mar 02	13:18–14:00	28°09,46′ N 14°10,29′ W	28°09,23′ N 14°07,76′ W	19–31	40–77
G10	13 Mar 02	16:18–17:20	28°07,78′ N 14°06,21′ W	28°08,34′ N 14°03,69′ W	429–572	959–1198
G11	14 Mar 02	10:05–10:40	28°07,30′ N 14°01,00′ W	28°07,29′ N 14°03,49′ W	31–140	1246–1347
G12	14 Mar 02	14:01–15:05	28°04,74′ N 14°01,90′ W	28°03,75′ N 13°59,02′ W	323–570	1406–1447
G13	14 Mar 02	16:56–18:20	28°04,17′ N 14°00,61′ W	28°02,22′ N 13°57,74′ W	632–1035	1447–1529
G14	14 Mar 02	21:30–22:00	28°08,76′ N 14°12,22′ W	28°07,43′ N 14°13,26′ W	15–43	96–211
G15	15 Mar 02	10:55–11:30	28°04,08′ N 14°09,47′ W	28°02,02′ N 14°11,03′ W	30–99	1439–1677
G16	15 Mar 02	12:52–13:30	28°03,70′ N 14°09,75′ W	28°01,63′ N 14°11,07′ W	185–211	1461–1680
G17	15 Mar 02	15:06–15:45	28°02,72′ N 14°10,50′ W	28°00,98′ N 14°12,46′ W	311–378	1000–1690
G18	15 Mar 02	16:50–17:45	28°03,37′ N 14°09,79′ W	28°05,29′ N 14°08,10′ W	470–695	1364–1491
G19	15 Mar 02	21:09–21:45	28°03,69′ N 14°09,38′ W	28°01,49′ N 14°11,04′ W	30–139	1462–1672
G20	16 Mar 02	13:18–14:35	28°03,42′ N 14°15,01′ W	28°00,26′ N 14°16,79′ W	308–498	1262–1592
G21	16 Mar 02	20:10–20:52	28°11,42′ N 13°59,82′ W	28°09,35′ N 14°01,68′ W	15–33	43–570
G22	16 Mar 02	22:14–22:50	28°03,20′ N 14°10,10′ W	28°01,27′ N 14°11,37′ W	215–261	1499–1701
G23	17 Mar 02	00:24–01:05	28°03,08′ N 14°10,24′ W	28°00,83′ N 14°11,73′ W	310–427	1505–1750
G24	17 Mar 02	02:47–03:40	28°02,86′ N 14°10,33′ W	28°00,79′ N 14°11,80′ W	565–980	1523–1750
G25	17 Mar 02	19:56–20:48	28°10,48′ N 14°02,56′ W	28°09,90′ N 13°59,80′ W	20–33	51–529
G26	17 Mar 02	22:03–23:35	28°04,36′ N 13°58,91′ W	28°03,31′ N 14°03,46′ W	25–62	1374–1577
G27	18 Mar 02	00:46–02:00	28°04,56′ N 13°59,73′ W	28°04,00′ N 14°03,56′ W	310–525	1364–1548

A commercial trawl with 80 m maximum horizontal opening was used (average vertical opening during tows 6.6 to 16.5 m). Cod-end mesh size was 2 mm and the opening of the cod-end was strengthened by a steel ring of 1.5 m diameter. Samples were scaled to a trawling time of 1 h. Characteristics of the vessel and the net, as well as a description of the fishing operations are given in Bordes et al. (1999) and Wienerroither (2003). The fishes were fixed in 7% formalin and later transferred to 70% ethanol. The specimens of the surveys ‘La Bocaina 11/97’ and ‘ECOS 04/99’ have been included into the collections of the Zoological Museum of the University of Copenhagen (ZMUC), Denmark, and of the Instituto Canario de Ciencias Marinas (ICCM), Telde, Gran Canaria, Spain. Specimens of the remaining surveys are all deposited at ICCM. Species identification is based on the following literature: Badcock (1982), Bigelow et al. (1964), Hulley (1981), Johnson (1974), Krefft (1970, 1971), Nafpaktitis et al. (1977), Nakamura & Parin (1993), Nelson (2006), Nielsen & Smith (1978), Nielsen & Bertelsen (1985), Rofen (1966), and Whitehead et al. (1984–1986). The occurrence of species in Canarian waters and general distribution have been checked using Badcock (1970), Badcock & Merrett (1976), Bordes et al. (1999), Brito et al. (2002), Hureau & Monod (1979), Kotthaus (1972), Lloris et al. (1991), Quéro et al. (1990), Rodríguez (2000), Rudyakov (1979), Whitehead et al. (1984–1986), and FishBase (http://www.fishbase.org).

Multivariate community analysis

Epipelagic trawl stations conducted during the day (i.e. G15 and G16) as well as singletons with respect to the whole data set (station B8) were excluded a priori. Hierarchical agglomerative clustering with group-average linking, based on the Bray–Curtis Similarity measure, was used to delineate groups with distinct community structure (Field et al. 1982). The normal or q-type analysis uses all species and treats samples as individual observations, each being composed of a number of attributes (i.e. the various taxa collected per tow). The analysis was done based on root–root transformed densities using the PC-ORD software package. The transformation down-weights the more abundant species and is invariant to scale range (Field et al. 1982). The specification of the similarity level for cluster separation was chosen to be data-driven. To compensate for possible misinterpretations derived from the exclusive use of dendrograms in community studies (Field et al. 1982), non-metric multidimensional scaling (MDS) was adopted. This method produces an ordination of the stations or species in two dimensions. Indicator species values were calculated using the method of Dufrene & Legendre (1997) to detect and describe the value of different species for the clusters derived by the classification and ordination method (McCune & Grace 2002). Information on the concentration of species abundance in a particular group is combined with the fidelity of occurrence of a species in a particular group, whereupon perfect indication means that the presence of a species points to a particular group without error, at least with the data set at hand. The statistical significance is evaluated by a randomization test (Monte Carlo Test, 1000 runs) which allows estimating exact significance without relying on the assumptions required for the standard asymptotic method.

Ecological diversity

The diversity and abundance of species was analysed using a representative set of indices (Magurran 2004) including total number of species and total number of individuals, species richness (Shannon–Wiener H’ and Simpsons D Index), evenness (Equitability J), and dominance (Berger–Parker). The hypotheses were tested that there are differences in diversity among (1) the principal clusters obtained from the multivariate analysis (result-driven data selection) and (2) areas within the Canarian archipelago (a-priori data selection). For comparison among areas epi- and mesopelagic tows south of Gran Canaria (epipelagic stations: B15, C12, F 17; mesopelagic stations: B14, C14, E5, F8, F9, F10, F14, F15, F18, F19) and south of Fuerteventura (epipelagic stations: B1, B3, B13, C4, C5, C6, C7, C8, G2, G4, G5, G7, G15, G16, G19, G21, G25, G26; mesopelagic stations: B2, G3, G6, G8, G10, G12, G13, G17, G18, G20, G22, G23, G24, G27), were grouped. Mean and confidence intervals for indices of clusters and areas were calculated and pairwise t-tests were used for additional statistical examination.

Beta diversity investigates the degree of association or similarity of sites or samples and is higher, the fewer species the different communities share (Magurran 2004). It was used to find possible differences (1) between the areas south of Gran Canaria and Fuerteventura and (2) variations among three selected groups of mesopelagic tows from survey 03/02 off SE Fuerteventura (Figure 2b; group 1: G6, G8, G10; group 2: G22, G23, G24; group 3: G12, G13; G27) to study small-scale discontinuities. Two indices, recorded to be the best to determine the degree of turnover in species composition, Whittaker β_W and Wilson & Shmida β_T (Wilson & Shmida 1984; Magurran 2004), were used based on presence and absence data.

Results

From a total of 104 trawl tows (35 coastal, 32 epipelagic and 37 mesopelagic), 93 resulted in the capture of 65,815 adult or juvenile fishes (14,609 epipelagic, and 51,206 meso-, bathy-, benthopelagic, or bathydemersal fishes). These belong to 19 orders, 51 families, 123 genera, and 211 species. Thirteen species are new records for the area. An overview of all species, indicating number of individuals and the survey in which they were captured is provided in Table II.

Table II. List of species encountered in the six surveys, with indication of number of specimens per survey. New records for the area are emphasized in bold, systematic order is according to Nelson (2006).

Order	Family/Sub-Family	Species	11/97	04/99	05/99	01/00	11/00	03/02
Squaliformes	Dalatiidae	Etmopterus pusillus (Lowe, 1839)					1
		Squaliolus laticaudus Smith & Radcliffe, 1912				1
Albuliformes	Halosauridae	Aldrovandia affinis (Guenther, 1877)					1
Anguilliformes	Derichthyidae	Derichthys serpentinus Gill, 1884						1
	Nemichthyidae	Avocettina infans (Guenther, 1878)	1		2		2	4
		Nemichthys curvirostris (Stroemman, 1896)	1	5			1	1
		Nemichthys scolopaceus Richardson, 1848	1	3		5	7
	Serrivomeridae	Serrivomer beanii Gill & Ryder, 1883	62	66	34	37	256	100
		Serrivomer lanceolatoides (Schmidt, 1916)	3	1	4	1	111	17
Saccopharyngif.	Saccopharyngidae	Saccopharynx ampullaceus (Harwood, 1827)	1				1	1
	Eurypharyngidae	Eurypharynx pelecanoides Vaillant, 1882					23	13
Clupeiformes	Engraulidae	Engraulis encrasicolus (Linnaeus, 1758)		6354			12	1
	Clupeidae	Sardina pilchardus (Walbaum, 1792)		43		1	4	3
		Sardinella aurita Valenciennes, 1847				35	24	160
		Sardinella maderensis (Lowe, 1838)		33			4	1
Argentiniformes	Opisthoproctidae	Dolichopteryx longipes (Vaillant, 1888)						1
		Opisthoproctus grimaldii Zugmayer, 1911						2
		Opisthoproctus soleatus Vaillant, 1888	1			1	7	2
	Microstomatidae	Bathylagus greyae Cohen, 1958		3	1	3		26
		Dolicholagus longirostris (Maul, 1948)						2
		Melanolagus bericoides (Borodin, 1929)			3			1
		Nansenia groenlandica (Reinhardt, 1840)						3
	Platytroctidae	Barbantus curvifrons (Roule & Angel, 1931)					4
		Holtbyrnia macrops Maul, 1957					1	12
		Maulisia mauli Parr, 1960						1
		Searsia koefoedi Parr, 1937	2				8	8
	Alepocephalidae	Bathytroctes microlepis Guenther, 1878		1
		Rouleina maderensis Maul, 1948						1
		Xenodermichthys copei (Gill, 1884)	1	1		2	8	8
Stomiiformes	Diplophidae	Diplophos taenia Günther, 1873	13	2	4		1	1
		Manducus maderensis (Johnson, 1890)		1
	Gonostomatidae	Bonapartia pedaliota Goode & Bean, 1896	1	3	3		3	6
		Cyclothone braueri Jespersen & Tåning, 1926	1281	586	804	954	3270	2569
		Cyclothone livida Brauer, 1902		5		2
		Cyclothone microdon (Guenther, 1978)					682	788
		Cyclothone pallida Brauer, 1902	4	9		11
		Cyclothone pseudopallida Mukhacheva, 1964	34	113	34	23	820	235
		Gonostoma denudatum Rafinesque, 1810	11	1	3	8	6	40
		Margrethia obtusirostra Jespersen & Tåning, 1919	5	5	2	1	3	46
		Sigmops bathyphilus (Vaillant, 1884)		2				1
		Sigmops elongatus (Guenther, 1878)	8	30	7	8	17	94
	Sternoptychidae	Argyropelecus aculeatus Valenciennes, 1850	2	4	5	5	5	9
		Argyropelecus affinis Garman, 1899			3	1
		Argyropelecus gigas Norman, 1930	3	3	6	7	26	15
		Argyropelecus hemigymnus Cocco, 1829	41	23	44	52	32	387
		Maurolicus muelleri (Gmelin, 1789)		5
		Sternoptyx diaphana Hermann, 1781	15	7	2	1	160	53
		Sternoptyx pseudobscura Baird, 1971					5
		Valenciennellus tripunctulatus (Esmark, 1871)	8	4	1	4	4	10
	Phosichthyidae	Ichthyococcus ovatus (Cocco, 1838)		2	1	2	6	19
		Vinciguerria attenuata (Cocco, 1838)	58	71	6	749	136	425
		Vinciguerria nimbaria (Jordan & Williams, 1895)	1328	260	674	59	4	97
		Vinciguerria poweriae (Cocco, 1838)	57	22	103	70	6	7
	Astronesthinae	Astronesthes gemmifer Goode & Bean, 1896		1	2		5	2
		Astronesthes indicus Brauer, 1902			1		1	3
		Astronesthes leucopogon Regan & Trewavas, 1929	1				1	1
		Astronesthes macropogon Goodyear & Gibbs, 1970				1	2	1
		Astronesthes micropogon Goodyear & Gibbs, 1970			1		2	5
		Astronesthes neopogon Regan & Trewavas, 1929						2
		Astronesthes niger Richardson, 1845			2
		Borostomias antarcticus (Loennberg, 1905)			1		1
		Borostomias elucens (Brauer, 1906)					1
		Borostomias mononema (Regan & Trewavas, 1929)		4		3		1
		Neonesthes capensis (Gilchrist & von Bonde, 1924)						1
		Rhadinesthes decimus (Zugmayer, 1911)		5		1		2
	Stomiinae	Stomias boa (Risso, 1810)	8	137	72	22	15	96
		Stomias brevibarbatus Ege, 1918		1	1		3
		Stomias longibarbatus (Brauer, 1902)	3	4		1		1
		Chauliodus danae Regan & Trewavas, 1929	58	196	88	67	288	91
		Chauliodus sloani Bloch & Schneider, 1801	17	34	30	66	69	63
	Melanostomiinae	Bathophilus brevis Regan & Trewavas, 1930		1
		Bathophilus digitatus (Welsh, 1923)		1	1	1	2	2
		Bathophilus longipinnis (Pappenheim, 1914)	2
		Bathophilus pawneei Parr, 1927		1	1
		Bathophilus vaillanti (Zugmayer, 1911)	14	23	8	3	13	24
		Chirostomias pliopterus Regan & Trewavas, 1930			4		5
		Echiostoma barbatum Lowe, 1843			7
		Eustomias acinosus Regan & Trewavas, 1930			1
		Eustomias bigelowi Welsh, 1923		1
		Eustomias braueri Zugmayer, 1911			1
		Eustomias fissibarbis (Pappenheim, 1912)			1
		Eustomias longibarba Parr, 1927	2				1
		Eustomias melanostigma Regan & Trewavas, 1930					2
		Eustomias obscurus Vaillant, 1884	39	16	2		17	4
		Eustomias schmidti Regan & Trewavas, 1930		1
		Eustomias simplex Regan & Trewavas, 1930		1
		Eustomias tetranema Zugmayer, 1913			2		1
		Flagellostomias boureei (Zugmayer, 1913)		1
		Leptostomias bilobatus (Koefoed, 1956)			1
		Leptostomias gladiator (Zugmayer, 1911)	1				2	2
		Leptostomias longibarba Regan & Trewavas, 1930			1			1
		Melanostomias bartonbeani Parr, 1927	1
		Melanostomias biseriatus Regan & Trewavas, 1930		2	1
		Melanostomias tentaculatus (Regan & Trewavas, 1930)		1			5	1
		Photonectes braueri (Zugmayer, 1913)		3	2			1
		Photonectes dinema Regan & Trewavas, 1930			6
		Photonectes margarita (Goode & Bean, 1896)		1
		Photonectes parvimanus Regan & Trewavas, 1930					3
	Idiacanthinae	Idiacanthus fasciola Peters, 1877	64	115	62	24	40	29
	Malacosteinae	Malacosteus niger Ayres, 1848					3	3
		Photostomias guernei Collett, 1889	43	51	5	1	16	44
Aulopiformes	Synodontidae	Synodus synodus (Linnaeus, 1758)	1
	Notosudidae	Scopelosaurus lepidus (Krefft & Maul, 1955)		2	1
	Scopelarchidae	Benthalbella infans Zugmayer, 1911					5	1
		Rosenblattichthys hubbsi Johnson, 1974			1			1
		Scopelarchus analis (Brauer, 1902)						2
	Evermannellidae	Evermannella melanoderma Parr, 1928	2				3	1
	Alepisauridae	Alepisaurus ferox Lowe, 1833					1
		Omosudis lowii Guenther, 1887						2
	Paralepididae	Anotopterus pharao Zugmayer, 1911					1
		Arctozenus risso (Bonaparte, 1840)					90
		Lestidiops affinis (Ege, 1930)	6
		Lestidiops jayakari (Boulenger, 1889)	25				1
		Lestidiops sphyrenoides (Risso, 1820)		1				2
		Macroparalepis affinis Ege, 1933						3
		Macroparalepis brevis Ege, 1933						1
		Magnisudis atlantica (Kroeyer, 1868)						22
		Paralepis brevirostris (Parr, 1928)					1	2
		Paralepis coregonoides Risso, 1820	1
		Stemonosudis intermedia (Ege, 1933)	2					2
		Sudis hyalina Rafinesque, 1810	2					2
Myctophiformes	Myctophidae	Benthosema suborbitale (Gilbert, 1913)	11	20	166	7	94	896
		Bolinichthys indicus (Nafpaktitis & Nafpaktitis, 1969)	12	293	48	84	295	139
		Bolinichthys supralateralis (Parr, 1928)		1			1	2
		Centrobranchus nigroocellatus (Guenther, 1873)		1
		Ceratoscopelus maderensis (Lowe, 1839)	61	24	2	16	177	612
		Ceratoscopelus warmingii (Luetken, 1892)	158	153	47	39	1544	170
		Diaphus adenomus Gilbert, 1905	132	2		1	2	1
		Diaphus dumerilii (Bleeker, 1856)	41	24		37	2584	23
		Diaphus effulgens (Goode & Bean, 1896)			1			1
		Diaphus holti Tåning, 1918		3	1	3	7	201
		Diaphus lucidus (Goode & Bean, 1896)			1			3
		Diaphus metopoclampus (Cocco, 1829)	1	15	12	9		18
		Diaphus mollis Tåning, 1928	72	24	29	65	645	32
		Diaphus perspicillatus (Ogilby, 1898)	24	1	3	15	53	43
		Diaphus problematicus Parr, 1928						1
		Diaphus rafinesquii (Cocco, 1838)	19	65	20	99	14	355
		Diaphus splendidus (Brauer, 1904)					1
		Diaphus termophilus Tåning, 1928	2	17		68	14	1
		Diaphus vanhoeffeni (Brauer, 1906)	1				2
		Diogenichthys atlanticus (Tåning, 1928)	23	247	145	12	125	207
		Gonichthys cocco (Cocco, 1829)		5		1	4	21
		Hygophum benoiti (Cocco, 1838)	336	1	11	13	5	25
		Hygophum hygomii (Luetken, 1892)	372	1462	1115	97	2320	1332
		Hygophum reinhardtii (Luetken, 1892)	32	143	138	15	45	134
		Hygophum taaningi Becker, 1965	44	56		3	23	96
		Lampadena chavesi Collett, 1905		3	1	1	7	4
		Lampadena speculigera Goode & Bean, 1896		1	1
		Lampadena urophaos Paxton, 1963		10	3		3
		Lampanyctus alatus Goode & Bean, 1896	2	78	56	30	109	464
		Lampanyctus crocodilus (Risso, 1810)		1	2	2	1	5
		Lampanyctus festivus Tåning, 1928		4	12	2	15	24
		Lampanyctus nobilis Tåning, 1928		2
		Lampanyctus photonotus Parr, 1928		16	12		5	20
		Lampanyctus pusillus (Johnson, 1890)	6	114	137	299	267	634
		Lepidophanes gaussi (Brauer, 1906)	35	113	106	56	184	189
		Lepidophanes guentheri (Goode & Bean, 1896)						1
		Lobianchia dofleini (Zugmayer, 1911)	955	364	1517	294	1144	1772
		Lobianchia gemellarii (Cocco, 1838)	52	29	51	118	34	156
		Myctophum nitidulum Garman, 1899	13	6	1	10	22
		Myctophum punctatum Rafinesque, 1810		107	4			14
		Myctophum selenops Tåning, 1928	1	14	1	3	13	12
		Nannobrachium atrum (Tåning, 1928)	4	56	52	2	16	184
		Nannobrachium cuprarium (Tåning, 1928)		6	2		3	5
		Nannobrachium lineatum (Tåning, 1928)		1			3
		Notolychnus valdiviae (Brauer, 1904)			1		29	31
		Notoscopelus bolini Nafpaktitis, 1975		10	4			8
		Notoscopelus caudispinosus (Johnson, 1863)			7	2		2
		Notoscopelus elongatus (Costa, 1844)						5
		Notoscopelus resplendens (Richardson, 1845)	63	86	290	11	168	335
		Symbolophorus veranyi (Moreau, 1888)	5	21	32	4	6	17
		Taaningichthys bathyphilus (Tåning, 1928)					4
		Taaningichthys minimus (Tåning, 1928)	1	3				1
Lampriformes	Stylephoridae	Stylephorus chordatus Shaw, 1791						1
	Regalecidae	Regalecus glesne Ascanius, 1772		1		1		1
Gadiformes	Melanonidae	Melanonus zugmayeri Norman, 1930						1
Lophiiformes	Melanocetidae	Melanocetus johnsonii Guenther, 1864						1
	Oneirodidae	Oneirodes anisacanthus (Regan, 1925)					1	1
	Ceratiidae	Ceratias holboelli Kroeyer, 1845	1				2	1
	Lynophrynidae	Haplophryne mollis (Brauer, 1902)					1
Beloniformes	Scomberesocidae	Scomberesox saurus (Walbaum, 1792)						1
		Scomberesox simulans (Hubbs & Wisner, 1980)					1
Stephanoberycif.	Melamphaidae	Melamphaes typhlops (Lowe, 1843)	2	5	10	1	112	11
		Poromitra capito Goode & Bean, 1883	3		1	1		2
		Poromitra megalops (Luetken, 1878)						3
		Scopelogadus beanii (Guenther, 1887)	1					11
	Rondeletiidae	Rondeletia loricata Abe & Hotta, 1963						1
	Cetomimidae	Cetomimus hempeli Maul, 1969					1
		Cetostoma regani Zugmayer, 1914					1	1
Beryciformes	Anoplogasteridae	Anoplogaster cornuta (Valenciennes, 1833)	1		2			1
	Diretmidae	Diretmus argenteus Johnson, 1864	8	9	6	4	2	7
Zeiformes	Grammicolepididae	Grammicolepis brachiusculus Poey, 1873				1
Gasterosteiformes	Macroramphosidae	Macroramphosus scolopax (Linnaeus, 1758)		16	1
Scorpaeniformes	Scorpaenidae	Setarches guentheri Johnson, 1862	1				1
Perciformes	Percichthyidae	Howella brodiei Ogilby, 1899					2
	Carangidae	Trachurus picturatus (Bowdich, 1825)	96	482		183	47	28
		Trachurus trachurus (Linnaeus, 1758)		428				122
	Caristiidae	Platyberyx opalescens Zugmayer, 1911		4	2		1
	Sparidae	Boops boops (Linnaeus, 1758)		195		1401	870	3
		Pagellus acarne (Risso, 1827)		17
		Pagellus erythrinus (Linnaeus, 1758)		11
	Chiasmodontidae	Pseudoscopelus altipinnis Parr, 1933	1
	Gempylidae	Diplospinus multistriatus Maul, 1948	116	13	17	2	9	5
		Nealotus tripes Johnson, 1865	4
		Promethichthys prometheus (Cuvier, 1832)					6
	Trichiuridae	Benthodesmus simonyi (Steindachner, 1891)		11		4
		Lepidopus caudatus (Euphrasen, 1788)		5	5	35		22
	Scombridae	Auxis rochei (Risso, 1810)	44
		Scomber colias Gmelin, 1789	106	2022	1	385	324	1166
	Nomeidae	Cubiceps gracilis (Lowe, 1843)		5	2	8
	Caproidae	Capros aper (Linnaeus, 1758)		6	1

Multivariate community analysis

The dendrogram resulting from the Bray–Curtis analysis of the 90 trawl stations is shown in Figure 3. Most of the stations are allocatable to four main clusters (1–4). One smaller cluster (cluster 5) is formed by four trawl stations. Two stations deviate in species composition from all others. The results of multidimensional scaling (MDS) using the same similarity matrix as above, delineating groups of stations from the dendrogram is shown in Figure 4. This analysis gives essentially the same picture as the dendrogram. The principal factor of association was the trawling depth in combination with the bottom depth. Mesopelagic tows are found in cluster 2 and the smaller cluster 5, whereas coastal tows are represented in cluster 3 and 4. Epipelagic oceanic tows formed the cluster 1 and are distributed in all other clusters except cluster 5. On account of the many stations no fixed similarity level for cluster separation was defined, therefore clusters vary from 45.8% (cluster 2) to 14.2% (cluster 4) similarity. Figure 5 illustrates epipelagic and coastal tows around Gran Canaria and Fuerteventura and their affiliation to the clusters, showing also the spatial overlap between cluster 3 and 4. Below is a detailed characterization of each of the five clusters.

• Cluster 1, similarity level 41.1%, 13 stations, mean water depth 1427 m, habitat epipelagic–oceanic: it comprises only epipelagic tows, characterized by a high variability in abundance of vertically migrating mesopelagic species and by three epipelagic species (Sardinella aurita, Scomber colias, Trachurus picturatus). Indicator species for this group are myctophids (Table III), known for their intense daily vertical migration behaviour. All tows are restricted to the oceanic realm (Figure 5).

• Cluster 2, similarity level 45.8%, 33 stations, mean water depth 1388 m, habitat mesopelagic: with the exception of one far-offshore epipelagic trawl station (D6), cluster 2 is composed of mesopelagic tows exclusively. The relatively high similarity (32.3%) of clusters 1 and 2 indicates a close relationship among the stations of these groups. In fact they share many vertically migrating species, but not the non- or little migratory species (found only in the mesopelagic tows), which delineate the two clusters. MDS (Figure 4) plots the stations of clusters 1 and 2 very closely, but also indicates the high similarity of the stations within and between the clusters. Non-migrating species like Serrivomer beanii and Sigmops elongatus, among others, are the best indicators for this group (Table III).

• Cluster 3, similarity level 17.0%, 12 stations, mean water depth 327 m, habitat coastal: this cluster comprises six coastal and six epipelagic samples, composed of several epipelagic species, mainly in low numbers, and vertically migrating myctophids, partly rich in individuals. Diaphus dumerilii is the best indicator species. The mesopelagic species in this cluster obviously perform an intense horizontal migration from the oceanic realm towards the islands (Figure 5).

• Cluster 4, similarity level 14.2%, 26 stations, mean water depth 205 m, habitat coastal: the cluster includes most of the tows taken above the continental shelf (coastal) plus eight oceanic epipelagic tows. It is characterized by the highest number of epipelagic species and individuals. As opposed to cluster 3, mesopelagic fishes (in terms of species and individuals) were less frequent in these trawl stations. Among the epipelagic species caught during the six surveys, only Scomber colias and Trachurus picturatus showed up regularly and are indicator species.

• Cluster 5, similarity level 44.3%, 4 stations, mean water depth 1294 m, habitat mesopelagic: the mesopelagic tows in this small cluster have a strong affiliation to cluster 2, the stations are partly overlapping (Figure 4), but differ in a slightly lower number of species.

Figure 3.  Dendrogram showing classification of the 90 trawl stations, four main clusters (1–4) can be distinguished. Abundances of species were root–root transformed before comparing stations using the Bray–Curtis measure, and the dendrogram formed by group-average sorting. The distance axis indicates the distance between groups, not expressed as a simple distance measure, but as Wishart's (1969) Objective Function (which is a measure of information loss as agglomeration proceeds): Cluster 1 = 6.95, Cluster 2 = 6.40, Cluster 3 = 9.80, Cluster 4 = 10.13. The second axis is based on the same function, but is converted to a percentage of information remaining: Cluster 1 = 41.1%, Cluster 2 = 45.8%, Cluster 3 = 17.0%, Cluster 4 = 14.2%.

Figure 4.  Ordination of the 90 trawl stations in two dimensions using multidimensional scaling on the same similarity matrix as Figure 3, final stress = 14.99250.

Figure 5.  Epipelagic and oceanic tows around Gran Canaria and Fuerteventura, indicating their affiliation to clusters in classification (Figure 3) and ordination (Figure 4).

Table III. Indicator species for the four clusters (I–IV) derived by the classification and ordination method. Relative abundance (RA) of the species in the group of stations over the average abundance of that species in all stations (in % of perfect indication). Relative frequency (RF) of the species in the trawl stations of the group (in % of perfect indication). Indicator values (IV) range from 0 (no indication) to 100 (perfect indication) and are based on a combination of RA and RF. Monte Carlo test of significance of observed indicator value for the species in the particular group. Only species with an indicator value of more than 50 are listed.

		RA %	RF %	IV %	p
Cluster 1	Lepidophanes gaussi	65	100	65.5	0.001
	Diogenichthys atlanticus	68	92	63.2	0.001
	Benthosema suborbitale	64	92	58.7	0.001
	Hygophum reinhardtii	62	92	56.9	0.001
Cluster 2	Serrivomer beanii	79	91	72.2	0.001
	Sigmops elongatus	81	82	66.4	0.001
	Cyclothone pseudopallida	79	82	64.7	0.001
	Bolinichthys indicus	64	88	55.9	0.001
	Cyclothone braueri	59	94	55.0	0.001
	Melamphaes typhlops	100	55	54.5	0.004
Cluster 3	Diaphus dumerilii	57	92	52.4	0.006
Cluster 4	Scomber colias	70	92	64.8	0.001
	Trachurus picturatus	68	77	52.7	0.002

Ecological diversity

Alpha diversity

Only the four main clusters (1–4) derived by classification and ordination methods are considered in the diversity analysis (Figure 6). Clusters formed by stations close to the shelf (3 and 4) tend to be different from clustered stations from oceanic waters (1 and 2). Cluster 2 (mesopelagic tows) has the highest diversity and equitability and the lowest dominance. This is in contrast to cluster 4, which is much less diverse but shows a high level of dominance (with cluster 3).

The number of species (Figure 6a) is significantly different between all clusters (F = 153.36, p<0.001) except between clusters 3 and 4, whereas there are no significant differences in number of individuals between the clusters. Significant differences in diversity indices (H’ between clusters 1–3, 1–4, 2–3, and 2–4, F = 56.12, p<0.001; Simpson D between clusters 1–4, 2–3, and 2–4, F = 17.02, p<0.005), equitability (Evenness J between clusters 1–3, 1–4, 2–3, and 2–4, F = 56.14, p<0.001) and dominance (Berger–Parker between clusters 1–4, 2–3, and 2–4, F = 14.38, p<0.005) among the clusters were revealed by pairwise t-tests, which coincide with non-overlapping 99% confidence intervals (Figure 6b–d).

Figure 6.  Mean and 99% confidence intervals of diversity and evenness indices (a: Number of species, b: Shannon–Wiener H’, c: Evenness J, d: Berger–Parker Dominance) of the principal clusters (cf. classification and ordination).

Examinations based on a-priori data selection to detect differences between richness, diversity, equitability, and dominance between the patches of trawl stations to the south of Gran Canaria and Fuerteventura, epi- and mesopelagic tows were treated separately. Pairwise t-tests revealed no significant differences.

Beta diversity

To the south of Fuerteventura beta diversity is considerably higher compared to the south of Gran Canaria (Table IV). The two habitats to the south of Gran Canaria have approximately similar beta diversity, whereas there is a remarkable difference between the epi- and mesopelagic realms south of Fuerteventura. Groups 2 and 3 have similar beta diversity, whereas group 1 (closer to the coast and the upwelling region off Gran Tarajal) is higher in β_W and β_T (Table IV).

Table IV. Beta diversity, Whittaker β_W (=S/α) and Wilson & Shmida β _T (=[g(H) + l(H)]/2S _j ), of two different areas, habitats, and groups of mesopelagic trawl stations in the south of Fuerteventura. (S=total number of species recorded, α=average sample diversity, g(H) = number of species gained, l(H) = number of species lost, S _j =species richness of sample j).

		Whittaker β W	Wilson & Shmida β T
Epipelagic realm	Gran Canaria	1.263	1.605
	Fuerteventura	4.208	3.981
Mesopelagic realm	Gran Canaria	1.729	1.620
	Fuerteventura	2.550	2.442
Groups of mesopelagic trawls	1	1.105	0.737
	2	0.767	0.688
	3	0.774	0.690

Discussion

Multivariate community analysis

The dense ordination of mesopelagic trawl stations (cluster 2) reveals that these have the most closely related species composition. Although several tows contain very rare and unique species, the tows are primarily unified by the dominant non- or little migrating fishes (e.g. Serrivomer spp., Cyclothone spp.). The far offshore position and the deep sounding of the only epipelagic station within this cluster might contribute to the species composition resembling the mesopelagic tows.

Clusters 1 and 2 showed the closest relationships. Their tows were all made in the oceanic realm and contain a high diversity of myctophids, which is their unifying factor. Lepidophanes gaussi is the main indicator species for cluster 1 and all the other indicator species are lanternfishes too (e.g. Benthosema suborbitale, Ceratoscopelus warminigii, Diogenichthys atlanticus). These inhabit the epipelagic realm during night for foraging reasons and are often encountered at higher densities there than in mesopelagic tows. Possible explanations for this phenomenon might be the limited space in near-surface habitats leading to a denser aggregation of specimens and/or biotic or abiotic factors with effects on spatial preferences. The most abundant species in these clusters are all mesopelagic (mainly myctophids), only the epipelagic Scomber colias was encountered in comparable abundances. This confirms the unusual horizontal migration into oceanic waters of this normally shelf-dwelling species, as indicated by Bordes et al. (1999).

Myctophids are also the most influential group in cluster 3. Although Diaphus dumerilii is the best indicator species, other lanternfish species occur in high numbers in trawl tows of this cluster too (e.g. Hygophum hygomii, Lobianchia dofleini, Notoscopelus resplendens). But the latter are also very abundant in other tows and therefore less appropriate as indicators. Diaphus dumerilii shows its highest abundances in rather shallow depths indicating a certain degree of land association. Cluster 3 combines trawls in and close to the coastal zone. Uiblein & Bordes (1999) ascribe the occurrence of mesopelagic species in such shallow waters to the abrupt changeover of depths around islands of volcanic origin. Nevertheless, the findings of such high numbers (e.g. 1561 D. dumerilii in survey 11/00 tow 3, 43–90 m trawling depth; 888 H. hygomii in survey 11/00 tow 1, 55–90 m trawling depth), differ from the observations by Hulley (1986) and Hulley & Prosch (1987), who recorded in the southern Benguela upwelling region the stock of only one myctophid species above a bottom depth of less than 300 m. Several species of Diaphus have been found to show spatial affinities to steep slope bottoms and hence characterized as ‘pseudoceanic’ (Merrett 1985; Hulley & Lutjeharms 1989).

Cluster 4 shows the highest heterogeneity, deriving from several epipelagic species, especially the two indicator species Scomber colias and Trachurus picturatus. Surprising is the presence of the myctophid D. dumerilii in this cluster comprising mostly coastal tows. In tows 5, 6, and 7 of survey 11/00, the abundance of the lanternfish was comparatively high (54, 128, and 1075 individuals) and also Boops boops (393 and 271 individuals in tow 5 and 7) and S. colias (387 individuals in tow 5; all numbers extrapolated) showed relatively high numbers. These three trawl stations were typically coastal, although this habitat is naturally very narrow around islands of volcanic origin. As proposed by Uiblein & Bordes (1999), close interactions among coastal, oceanic, benthic, and benthopelagic organisms are induced by the abrupt topography with possible predation of epipelagic species on myctophids like D. dumerilii. Impingement of pseudoceanic species may lead to intense spatial and trophic interactions with slope community (Reid et al. 1991; Gordon et al. 1995; Uiblein & Bordes 1999). The more dispersed ordination of the stations of group IV compared to the other large clusters (Figure 4) reflects irregular occurrences of many species.

Clusters 1 and 2 are clearly defined by habitat (oceanic–epipelagic and mesopelagic), whereas there are spatial overlaps between tows affiliated to clusters 3 and 4 (Figure 5). Seasonal effects can be excluded, as coastal tows within surveys show affiliation to different clusters. This suggests that occurrence and abundance of species is locally and temporally limited to a rather small scale instead of showing a constant pattern throughout the study area. The indicator species are not necessarily restricted to a single cluster (Table III). Specifically D. dumerilii and S. colias were found in stations of all clusters (except S. colias in cluster 2). Although numbers of individuals vary somewhat significantly, this clearly shows an intensified use of coastal habitats by mesopelagic fishes as well as of oceanic habitats by coastal species, respectively.

Several authors used a similar approach to investigate mesopelagic nekton assemblages. Hulley & Lutjeharms (1989), for instance, found that in the southern Benguela region the depth of the water column was the reason for a change in the lanternfish fauna and distinguish between oceanic and pseudoceanic species. Depth is also the deciding factor in cephalopod assemblages along the Spanish Mediterranean coast (González & Sánchez 2002). Fock et al. (2004) indicate that habitat structure in conjunction with physical and biological features is an important determinant of community structure. Benoit-Bird & Au (2006) found myctophids as the only animals in shallow waters of Hawaiian Islands, migrating actively with a striking horizontal component. Furthermore, these myctophids were of larger size than those found in farther offshore, deeper layers, which they ascribed to differences in swimming speed.

Ecological diversity

Alpha diversity

The diversity indices used confirm the often-stated characteristics of the mesopelagic realm, reported to be the most diverse and densely inhabited area of the open oceans (National Research Council 1995). The high species richness and evenness, and the low dominance values obtained in the current study agree with this statement. Quite contrasting are the mainly coastal tows of cluster 4. These stations in shallow bottom depths are characterised by low species richness, an often high number of individuals (and biomass) and one or two dominating epipelagic species. Clusters 1 and 3 are positioned in between. Cluster 1 (epipelagic tows) is in its diversity closer to cluster 2, with high richness and evenness but low dominance. This can be ascribed to the vertical migrants represented in these tows by high numbers (in species and individuals) and the lack of the partly rare non-migrators. Cluster 3 resembles cluster 4, although these tows are a special case: made close to the coast (some of them even coastal), with mesopelagic fishes, remarkably high in individuals but low in species.

Beta diversity

Depths of more than 2000 m are encountered close to the investigation area of Gran Canaria, whereas this depth zone is farther away from that of Fuerteventura. The differences in diversity might reflect the prevailing depth zones in the respective areas or general differences in spatial heterogeneity. The higher oceanic influence and the eddy south of Gran Canaria (Arístegui et al. 1994, 1997; Bordes et al. 1999) may have an effect on species distribution. Arístegui et al. (1989) found high primary production and large populations of mesozooplankton in coastal waters south of Gran Canaria compared to oceanic waters. Barange et al. (1998) emphasized the intense influences of hydrological features on a pelagic community in general. Even more remarkable is the distinctness between the epi- and mesopelagic habitat off south Fuerteventura. It emphasizes the complex connection of hydrological features and species distribution. Uiblein et al. (1996, 1998) discussed the existence of a seasonally limited local upwelling off Gran Tarajal, southeast Fuerteventura. Within the sub-area groupings of mesopelagic tows the two groups farther away from the coast have similar beta diversity, that clearly differs from group 1 indicating a heterogeneous species distribution at the scale of a few kilometres distance only. The mesopelagic boundary community of Hawaiian Islands shows similar regional differences in taxonomic composition, abundance and diversity (Reid et al. 1991; Benoit-Bird et al. 2001; Lammers et al. 2004). Benoit-Bird & Au (2006) refer the fine scale separation of habitat use to a reduction of competition between individuals in these dense aggregations relative to night time. Heterogeneity in species distribution depending on topography and hydrography seems to be a common feature, as it was also detected between adjacent deep-sea canyons off the Georges Bank, NW Atlantic (Uiblein et al. unpublished data).

This study has resulted in the collection of 13 new pelagic fish species for the Canary Islands. Not all islands and habitats have been covered by the surveys. We have presented here clear evidence of considerable variation in local species composition reflecting the influences of topography and hydrography and the interactions between these two variables. Both epipelagic and mesopelagic fishes seem to respond actively to physical heterogeneity at the scale of a few kilometres only, thus inducing patchiness and turnover among close-by areas. The high number of species encountered, the still ongoing discovery of hitherto unknown species, and the finding of distinct, taxon-specific distribution patterns strongly suggest that pelagic fishes are an important component for the assessment and functional understanding of biodiversity around oceanic islands.

Editorial responsibility: Peter R. Møller

Acknowledgements

All surveys were carried out in projects supported by the Viceconsejería de Pesca del Gobierno de Canarias. Rupert Wienerroither got financial support from the ‘Stiftungs- und Förderungsgesellschaft der Paris-Lodron Universität Salzburg’. Sincere thanks to all colleagues at the Zoological Museum Copenhagen (ZMUC) and the University of Salzburg, especially A. Lametschwandtner and J. Strobl.

Notes

Published in collaboration with the University of Bergen and the Institute of Marine Research, Norway, and the Marine Biological Laboratory, University of Copenhagen, Denmark