Abstract
Invertebrate primary succession is investigated across a chronosequence in the subalpine (birch woodland) zone at Fåbergstølsbreen using pitfall traps. Presence and abundance of taxa, functional groups and communities are examined using a geo-ecological approach including mapping, graphical analysis, two-way indicator species analysis (TWINSPAN) and non-metric multidimensional scaling (NMS). Twenty-nine of the 67 recorded taxa, and 14 of the 37 epigeal (surface active) taxa colonize terrain deglacierized for <20 years and catches of most of these pioneer taxa attain an early peak on terrain deglacierized for <40 years. Catches of most later colonizers peak in the mature (‘climax’) stage where 86% of the pioneer taxa and 79% of the epigeal pioneers are also recorded. The number of taxa increases across the chronosequence as new taxa, predominantly predators, appear but relatively few taxa drop out of the succession, and as the dominant species in the traps changes from the harvestman, Mitopus morio, to the wood ant, Formica lugubris. The invertebrate communities are interpreted as loosely organized groupings of individualistic species, the distributions of which are strongly related to terrain age. Changes during succession are interpreted as driven primarily by competition between the carnivorous taxa, though herbivores and omnivores comprise up to 24% and 11% of the taxa, respectively, in both pioneer and mature zones. Hence the ‘addition and persistence model’ of invertebrate succession, previously proposed for the alpine zone, is largely substantiated but partially modified for the subalpine zone, where succession is generally more rapid, overall changes during succession are greater, communities are better organized, food webs appear better developed, and there is an element of replacement change in the later stages.
Keywords
Introduction
Since the pioneering research of Janetschek (1949) in the Austrian Alps, models of invertebrate succession on glacier forelands have been greatly influenced by supposed similarities with and dependence on vegetation succession. Vegetation succession on glacier forelands has been investigated in considerable detail in many different regions (see reviews in Matthews, 1992, 1999; Nagy and Grabherr, 2009; Walker and del Moral 2003). However, the patterns and processes of invertebrate succession are relatively poorly understood in these and other habitats. The dominant model of primary vegetation succession emphasizes replacement change: i.e. the sequential colonization of plant communities, with pioneer species being replaced later by distinctly different communities. Recent investigations of the succession of various groups of invertebrates have revealed some differences from this ‘classical model’, including the facts that some pioneer invertebrate species, especially predators, can establish themselves rapidly, ahead of plant colonizers (Gereben-Krenn et al., 2011; Gobbi et al., 2006; Hågvar, 2012; Hodkinson et al., 2001, 2002; Kaufmann, 2001), and that the succession of invertebrates may exhibit greater continuity than previously proposed (Kaufmann, 2001; Kaufmann et al., 2002).
Some studies of invertebrate succession on glacier forelands have often stressed that species turnover is conspicuous after the pioneer stage with little overlap in species composition between early and later successional stages (e.g. Janetschek, 1949; Kaufmann, 2001; Macfadyen, 1963), reflecting the influence of vegetation and soil development on invertebrate community development (Kaufmann and Raffl, 2002), the later establishment of herbivores and detritivores (Bardgett et al., 2005), and the development of functional links between soil-dwelling or epigeal (surface active) invertebrates and vegetation (Albrecht et al., 2010; Carlson et al., 2010; Gobbi et al., 2010). However, other results, especially from the high Arctic (Hodkinson et al., 2001, 2002, 2004) and the alpine zone in southern Norway (Bråten et al., 2012; Hågvar, 2010, 2012; Hågvar et al., 2009; Vater, 2006, 2012), provide little support for major vegetation effects on invertebrate succession. This led Vater (2012) to propose an alternative model of invertebrate succession that emphasizes the progressive addition and the persistence of taxa, rather than replacement change, with comparatively little plant–invertebrate interaction.
The broad aim of this paper is to test further the ‘addition and persistence model’ using a glacier-foreland chronosequence from the subalpine zone in southern Norway. There are three main objectives: first, to describe the succession of invertebrate taxa and the pattern of invertebrate community development at Fåbergstølsbreen, a southeastern outlet of the Jostedalsbreen ice cap (Figure 1); second, to compare the results with those found, using similar methods, in the alpine zone at Storbreen, Jotunheimen; and third, to discuss the broader implications of the results in the context of the theory of invertebrate succession on glacier forelands.
Location of the Fåbergstølsbreen glacier foreland, an eastern outlet glacier of the Jostedalsbreen ice cap, southern Norway.
Study site
Fåbergstølsbreen glacier foreland (61°43′N, 7°18′E) occupies an east-facing tributary valley of Jostedalen (Figure 1). It extends approximately 2 km from the glacier snout (approximately 620 m above sea level) to its lowest point at about 480 m a.s.l. The area is characterised by subalpine woodland dominated by the downy birch (Betula pubescens), which forms part of the northern boreal zone (Moen, 1999). Data from Bjørkhaug-i-Jostedalen meteorological station (324 m a.s.l.) shows a mean annual air temperature of +3.7°C (July mean +13.4°C; January mean −4.9°C) and a mean annual precipitation of 1380 m (Aune, 1993; Førland, 1993). Geologically, the area is composed of granitic to granodioritic gneiss (Lutro and Tveten, 1996; Sigmond et al., 1984).
The chronosequence is based on a well-developed system of ‘Little Ice Age’ moraines dated by glacier monitoring, aerial photographs, historical evidence and lichenometry (Bickerton and Matthews, 1993). The outermost moraine, which dates from the ‘Little Ice Age’ maximum of c.
The vegetation succession on glacier forelands in the subalpine birch belt has been described by Robbins and Matthews (2010), who recognize a herbaceous pioneer Poa-Oxyria community that can develop, under favourable habitat conditions, into birch woodland with a dwarf-shrub understorey within 70 years (see also Fægri, 1933). However, there is a variety of habitats in response to microtopography and disturbance processes, associated in particular with variable drainage conditions on moraine slopes and the shifting courses of glaciofluvial meltwater streams, respectively.
Methods
Sampling and analytical methods were similar to those used by Vater (2012). The glacier foreland was subdivided into five zones of increasing age, from Zone 1 (terrain age 0–20 years; deglacierized between
The age zones of the chronosequence at the time of fieldwork (2004).
A total of 90 pitfall traps were arranged according to a ‘three-trap system’ (i.e. three traps were located in close proximity at each site). This introduced an element of replication while also taking account of small-scale habitat variability (Vater, 2006, 2012). Progressively fewer sites were located in Zone 1 (seven sites) to Zones 4 and 5 (three sites each), and five sites were located in Zone 6: to allow for expected patterns of diversity and heterogeneity through time. The variable number of traps per time zone was not ideal as the number of taxa recorded will increase with the number of traps in use. However, tests made on the Storbreen glacier foreland showed that as few as three traps are sufficient to provide a reasonably representative sample of of the presence of taxa at single sites (Vater, 2006, 2012). Furthermore, number of taxa recorded in each of Zones 4 and 5 at Fåbergstølsbreen was close to the number recorded when results from these two zones were combined together (see below). Thus, the effects of different numbers of traps per zone are regarded as insignificant in this data set based on a minimum of nine traps per zone.
Each trap consisted of a plastic cup set into the ground with the 70 mm diameter rim set level with the surface. Raised, plastic, rain-proof and cone-shaped lids prevented the traps filling with water and protected against small mammals. A 30% solution of ethanol (25 ml) was used as the killing and preserving agent, and 5 ml of ethylene glycol were added to reduce evaporation. Two collections were carried out in July to early September in 2004. Each collection represented the invertebrates trapped over a period of 12 or 18 days, each trap being filled and emptied twice. As collections were made by visiting the traps sequentially and all traps were emptied and filled within a few days of each other, any phenological changes between the early and late collections are likely to affect all age zones of the chronosequence equally. The average trapping rate – number of individuals per site per day – over the two collections, was used in subsequent numerical analyses involving abundance/activity. Nomenclature follows Chinery (1993), Olsen (1999) and Roberts (1995) (see also Fauna Europea, www.faunaeur.org). Most epigeal taxa (86%) were identified to species level or represent single species, whereas most flying invertebrates were identified only to family level.
The catches in the pitfall traps measure abundance of the epigeal insects (especially beetles and ants) and arachnids (spiders, harvestmen and mites) in the form of surface activity (rather than density). Although several epigeal taxa are able to fly (e.g. several beetles), where they have regular non-flying surface activity they are classified as non-flying epigeal taxa. Flying insects, include Diptera (flies), Hemiptera (true bugs), most Hymenoptera (sawflies, bees and related taxa), Plecoptera (stoneflies) and Trichoptera (caddisflies). These were analysed even though pitfall traps are not designed for flying insects, the presence of which in the traps may sometimes represent only chance occurrences and the absence of which may not be indicative of absence in the field.
The distribution and abundance of all individual taxa from 29 sites were mapped in geographical space and analysed graphically in relation to age zones. In order to test the ‘addition and persistence model’, the timing of first and last appearance and of the attainment of maximum abundance/activity of each taxon in the chronosequence were analysed. This focused on whether replacement change of taxa occurs during succession. Possible bias of the estimated number of taxa due to the inclusion of the flying insects and family-level data was investigated by carrying out separate analyses of the epigeal taxa and the epigeal taxa identified to species level. This demonstrated that inclusion of flying taxa and family-level units had no appreciable effect on the results (see below).
Temporal patterns in functional groups (carnivores, herbivores and omnivores) were analysed in a similar way, again with separate consideration of the epigeal taxa. The diet of the adult life phase was used in the definition of the functional groups: carnivores included predatory, parasitic and scavenging carnivores; herbivores included phytophagous, mycetaphagous (fungal eating) and scavenging herbivores; and omnivores included predatory and scavenging taxa that may supplement their diet with vegetation or plant detritus (detritivores). All taxa apart from five unidentified species of Hemiptera (designated H4, H5, H9, H16 and H18) were assigned to a functional group.
Community analysis involved classification of all taxa using two-way indicator species analysis (TWINSPAN; Kent and Coker, 1992; Lepš and Šmilauer, 2003; Kent, 2012; van Tangeren, 1995), mapping and ordination using non-metric multidimensional scaling (NMS; Clarke 1993; Kent, 2012; Lepš and Šmilauer, 2003; McCune and Grace, 2002). Eight species groups and seven site groups were considered optimal in terms of retaining interpretable detail. Twelve rare species (occurring at <5% of the sites, i.e. at one site only, mostly in Zone 6) were excluded from the classification and ordination analyses to prevent them from affecting the statistical stability of the results.
Results
Distribution of taxa and functional groups
The average collection rates of the 67 taxa recorded in each of the six age zones are presented in Table 2, and dynamic aspects of these patterns are summarized in Figure 2a–d. Zone 1 (terrain age 0–20 years) is characterized by relatively high taxonomic diversity (29 taxa of which 14 are epigeal taxa) but relatively low abundances of each taxon (Figure 2a), none of which are exclusive to this zone (Table 2). Several epigeal taxa, which may be considered pioneer species (including Acidota crenata, Pardosa pullata and Amara alpina), reappear in older zones even though they did not occur in Zone 2 (Table 2).
Presence and abundance/activity of the trapped taxa across the six age zones of the chronosquence. Values are catch sizes (individuals per site per day). Darker shading indicates the zone in which the taxon attains maximum abundance (zone of peak catch). Functional groups: C: carnivore; H: herbivore; O: omnivore; n.a.: not available. Taxon codes are labels for reference specimens.
Patterns in the dynamics of taxa and functional groups across the chronosequence: (a) number of taxa and epigeal taxa per age zone; (b) number of first and last appearances of taxa and epigeal taxa per age zone; (c) number of taxa, epigeal taxa and flying taxa attaining peak abundance (maximum catches) per age zone; (d) number of carnivores, epigeal carnivores, herbivores and omnivores per age zone.
Despite an increase in the number of taxa with terrain age across the chronosequence (Figure 2a), the number of taxa in Zones 2–5 (31–39 taxa) is little higher than in Zone 1. The similar form of the curve for epigeal taxa alone, which is almost identical to that of the epigeal taxa identified to species level, should also be noted. However, it is possible that the taxon Staphylinidae (rove beetles) may hide a high number of epigeal species of unknown successional pattern.
In the oldest zone (Zone 6), the number of taxa and the number of epigeal taxa both reach distinct maxima of 55 and 30 taxa, respectively, which can be attributed to the number of first appearances exceeding the number of last appearances (Figure 2b). In contrast to the arrival of 3–9 new taxa (of which two-thirds are epigeal taxa) in each of Zones 2–5, there are relatively few last appearances as terrain age increases. Indeed, no taxa appear to drop out of the succession before Zone 3 (terrain age 41–64 years) and the number of last appearances associated with Zones 3–5 remains extremely low (2–6 taxa per zone, of which half are epigeal taxa). There are no last appearances shown in Zone 6 because there is no later stage.
A substantial number of taxa attain peak abundance on relatively young terrain (24 in Zones 1 and 2, of which 10 are epigeal taxa and 14 are flying taxa) but relatively few (14 in total) on terrain of intermediate age in Zones 3–5 (Figure 2c). However, about half of the recorded taxa (34) and over half of the epigeal taxa (20) reach their maximum abundance in Zone 6.
The distributions of carnivorous, herbivorous and omnivorous taxa across the chronosequence are summarized in Figure 2d. These functional groups show a similar pattern to that of the overall number of taxa with carnivores in the majority in all zones. The proportion of carnivores is high throughout, greatest in Zones 1–4 (65–72% of taxa), falling to 53% in Zone 6. Of the epigeal taxa, 78–94% are carnivores in Zones 1–4 and the proportion remains high in Zones 5 and 6 (81% and 73%, respectively). Though much lower in numbers throughout the chronosequence, the proportion of herbivores is relatively high in Zone 1 (24%), progressively less in Zones 2–4 (declining to <10% of taxa in Zone 4) before rising again to >20% in both Zones 5 and 6. Omnivores are comparatively rare (6–11% throughout) rising slightly in the older zones.
Distributions of seven epigeal taxa and one flying taxon are illustrated in Figure 3 while maps of the other taxa are available in Vater (2006). Amara alpina (Figure 3a) is one of 11 taxa that attain peak abundance in Zone 1, whereas A. quenseli (Figure 3b) and Mitopus morio (Figure 3c) are two of the 13 that peak in Zone 2). Both of the Amara beetles appear to be pioneers but also occur with lower abundance on older terrain, including sites in Zone 6. However, whereas A. alpina is absent from Zones 3 and 4, A. quenseli occurs in all zones. The harvestman, Mitopus morio, also colonizes rapidly and is the commonest species on relatively young terrain (Zones 1–3) but its abundance is much reduced in Zone 3 and in the older zones. The next three beetles colonize more slowly. Notiophilus aquaticus (Figure 3d) reaches maximum abundance in Zone 3: although it is less abundant in the older zones, this decline appears particularly patchy. Pardosa lugubris (Figure 3e) and P. trailli (Figure 3f), in common with other Pardosa species, reach their peak abundance in Zone 6. The wood ant, Formica lugubris (Figure 3h), which first appears in Zone 3, is the commonest species in Zones 4–6 (where it was found in all traps), increasing rapidly in abundance by Zone 5 and becoming extremely abundant in Zone 6. The final example is Bombus sp. (Figure 3g), a bumblebee that occurs occasionally in the two oldest zones, and represents the large number of species that were trapped only occasionally on the oldest terrain.
The distribution patterns of eight taxa on the Fåbergstølsbreen glacier foreland: (a) Amara alpina; (b) Amara quenseli; (c) Mitopus morio; (d) Notophilus aquaticus; (e) Pardosa lugubris; (f) Pardosa trailli; (g) Bombus sp.; (h) Formica lugubris. Circle size denotes abundance: units are individuals per site per day. Note the abundance scale differs for each taxon.
TWINSPAN species groups
The first division of the taxa by TWINSPAN separated 24 widely distributed taxa from 31 that occur predominantly on relatively old terrain. The second division of the former group split off 14 taxa that tend to favour relatively young terrain, while the second division of the latter split off seven taxa that are sparsely distributed on terrain of intermediate age (as well as occurring sparsely on relatively old terrain). After three divisions, TWINSPAN defined eight interpretable groups of species of which those shown in Figure 3 may be considered examples. It should be noted, however, that the members of each group exhibit considerable individuality in their distributions.
Species group 1 – non-persistent pioneers
Included are a small number of diverse taxa – Mitopus morio, Braconidae sp., Bembidion sp. and Nebria nivalis/N. rufescens – that are characteristic of the youngest terrain where M. morio (Figure 3c) is dominant and the other three taxa occur in low abundance. Only M. morio was recorded in either of the two oldest zones.
Species group 2 – persistent pioneers
This large group of ten pioneer taxa tend to be relatively abundant on young terrain ages and persist throughout the chronosequence. Although they occur on relatively old terrain, these taxa linger on in the older zones in much lower numbers. Most of these are flying taxa, Diptera (Muscidae spp., Nematocera spp., Heledromia spp., Mycetophila spp., and Delia sp./Zaphne frontana) or Hemiptera (H9 and H16); the others include three epigeal species, Amara quenseli (Figure 3b), Bembidion fellmanni, and Trombidium holosericeum.
Species group 3 – persistent pioneers of low abundance
This second large group of pioneers includes eight taxa, four epigeal species (Erigone longipalpis, Malthodes flowoguttatus, Miscodera arctica, Staphylinidae spp.) and four flying taxa (Belytinae spp., Ichneumonidae spp., Beckerina spp. and Drymeia spp.). Although they are generally of lower abundance than the taxa in species group 1 in the youngest zones, they are equally persistent on relatively old terrain.
Species group 4 – returning pioneers
The two taxa in this very small group of species – Amara alpina (Figure 3a) and Eudasyphora cyanicolour/Cynomya mortuorum – combine maximum abundance in Zone 1 with very few occurrences in Zones 3–5 and moderate abundance in Zone 6. Although other pioneer taxa often exhibit a resurgence in numbers on relatively old terrain, those of this group uniquely display a break in the pattern of persistence. In the case of E. cyanicolour/C. mortuorium, however, this could be explained by these two flying species having different distributions.
Species group 5 – mid-successional taxa
This second very small group of two species includes Notiophilus aquaticus (Figure 3d) and the dipteran, Bibio langerus, both of which are absent from Zone 1 and attain maximum abundance on terrain of intermediate age (although B. langerus occurs at only two sites).
Species group 6 – mid/late-successional taxa
The five taxa in this group – two epigeal species (Catops tristis and Xysticus cristatus) and three flying insects (Mycetaulus bipunctatus and two species of Hemiptera) – also characterize terrain of intermediate age but tend to appear first in Zone 3 rather than Zone 2.
Species group 7 – late-successional/‘climax’ taxa
This large group of eight taxa – Formica lugubris (Figure 3h; the dominant species of the older zones), four other epigeal taxa (Gonatium rubellum, Drusilla spp., Queduis sp. and undifferentiated and juvenile Pardosa spp.) and three flying taxa (Sphaerophoria rueppellii, Cecidomyidae sp. and Suillia spp.) – are found almost exclusively in Zones 4–6, usually in appreciable numbers and with peak numbers in Zone 6.
Species group 8 – ‘climax’ taxa
A very large group of 15 taxa occur mostly on the oldest terrain of Zone 6, sometimes almost exclusively so (hence the attribution of ‘climax’ status). This diverse group includes many relatively rare taxa, including Bombus sp. (Figure 3g). Other members of the group include ten epigean species – Pardosa lugubris (Figure 3e), P. trailli (Figure 3f), P. palustris, P. pullata, Pirata piraticus, Zelotes subterraneus, Otiorhynchus nodosus, Agroeca proxima, Acidota crenata – and five flying taxa (Tipula excisa, Capnia atra, Dolichopus discifer, Limnephilidae spp. and a species of Hemiptera).
TWINSPAN site groups
The spatial distribution of seven TWINSPAN site groups after three divisions is shown in Figure 4 and their distribution in relation to the age zones is summarized in Table 3. Although the site groups show a clear association with terrain age with very little overlap in space, they do not correspond one-to-one with the species groups described above. The first division of the sites by TWINSPAN separated the eight sites located in Zones 5 and 6 from the remaining 21 sites. The second division separated the smaller group into those sites within the glacier foreland (Zone 5) from those outside it (Zone 6), while the larger group was split into sites predominantly in Zones 1 and 2, and sites predominantly in Zones 3 and 4.
The distribution pattern of the seven TWINSPAN site groups on the Fåbergstølsbreen glacier foreland. Characteristic colours identify sites within the same site group.
Summary of site-group community dynamics over the chronosequence. The numbers of sites assigned to each TWINSPAN site group are shown within each age zone.
Site group 1 – early-pioneer sites. Site group 1 is exclusive to Zone 1 on terrain deglacierized within 20 years.
Site group 2 – late-pioneer sites. This group is most common after ~30 years but occurs in Zones 1–3 on terrain deglacierized for up to ~60 years.
Site group 3 – early-successional sites. Absent from Zone 1, this group is most common late in Zone 2 on terrain deglacierized for ~40 years and persists alongside site group 2 in Zone 3.
Site group 4 – mid-successional sites. Absent from both Sites 1 and 2, this group first occurs in Zone 4 after ~60 years but is most common in Zone 4 (deglacierized for up to ~150 years) where it is the only site group present.
Site group 5 – late-successional sites. This group occurs exclusively in Zone 5, deglacierized for ~150–250 years.
Site groups 6 and 7 – ‘climax’ sites. These groups occur exclusively outside the glacier foreland boundary on the oldest terrain (Zone 6). Group 6 is a single site that contains many similar taxa to site group 7 but also has similarities to site group 5.
NMS ordination
Further clarification of the nature of the site groups and their interrelationships is shown in the NMS biplot (Figure 5), which provides a visual representation of the similarities between sites in a two-dimensional ‘species space’. The stress statistic of 9%, which is a ‘badness-of-fit’ measure, indicates that this two-dimensional representation of complex, multidimensional relationships contains little distortion. The site groups generally occupy distinct areas of this space of reduced dimensionality and the temporal pattern of the chronosequence is clearly reflected in the curvilinear arrangement of sites. The pioneer and early-successional site groups 1–3 occupy the top left of the space, whereas the late successional and climax groups 5–7 occupy the bottom right.
Two-dimensional NMS plot of the seven TWINSPAN site groups (stress 9.144; stability 0.00001). Envelopes and characteristic colours identify the site groups. The thick arrow indicates the trajectory of succession (inferred from Figure 4) from pioneer to mature zones.
Most of the site groups show little or no overlap in the ordination, indicating distinctive combinations of taxa presence and abundance. Other important features include: the wide differences in invertebrate composition between the pioneer and early-successional groups; the relatively high within-group variability exhibited, especially by the pioneer and early-successional groups; the comparatively large overlap between groups 5 and 7 (indicating similarities in community composition on either side of the glacier–foreland boundary); and, finally, the particularly distinct nature of mid-successional group 4 occupying an intermediate position that is distant from both the early- and late-successional groups (indicating changes in composition continue to occur in mid succession).
Discussion
Nature of colonization and succession at Fåbergstølsbreen
Twenty-nine taxa and 14 epigeal taxa were recorded on terrain deglacierized for <20 years (Zone 1) where vegetation cover is generally low. This rapid initial colonization is in accord with many other studies, which have emphasized such factors as the mobility of many invertebrates, including the non-flying, epigeal taxa (Coulson et al., 2003), the ability of carnivores (such as carabid beetles, spiders and harvestmen) and detritivores to subsist in the absence of vegetation (Hågvar, 2012; Hodkinson et al., 2002; König et al., 2011) and the possibility of invertebrates feeding on heterotrophic micro-organisms that may utilize ancient carbon (Bardgett et al., 2007).
Taxa numbers, including the number of epigeal taxa, continue to rise relatively slowly on terrain deglacierized for 21–64 years (Zones 2 and 3), then level off in older Zones 4 and 5 (Figure 2a). Together with concurrent low numbers of first appearances after Zone 1 (Figure 2b), this suggests that most taxa able to survive the environmental conditions on recently deglacierized terrain arrive early in the succession and are not limited by their dispersal abilities (cf. Ingimarsdóttir et al., 2012). This may, in turn, partly reflect the relatively short distances to source populations beyond the glacier–foreland boundary, including the probable existence of populations of many of the taxa on the valley sides high above the foreland.
The appreciable number of taxa with peak catches in Zones 1 and 2 (Figure 2c) indicates that most of these early colonizers are able to reproduce and expand their populations soon after reaching the site, as suggested in earlier studies (e.g. Gereben-Krenn et al., 2011). However, the epigeal taxa recorded on the youngest terrain (Zone 1) build up their populations somewhat more slowly than the flying taxa. Whereas peak catches of many flying taxa are attained within 20 years, the epigeal taxa attain their early peaks only after 20–40 years (Zone 2; Figure 2c). Fewer taxa have peak catches in Zones 3–5, where competition may be reducing the abundance of many of the pioneer species whilst having a relatively small effect on the number of taxa. Particularly important as an underlying cause of this pattern is likely to be the increasing abundance of Formica lugubris, which first appears in Zone 3 and becomes extremely abundant on older terrain.
Lack of dependence of the invertebrate succession on the vegetation succession is supported by the fact that predatory carnivores predominate not only in the pioneer zone but also make up a high proportion of the invertebrate fauna throughout the chronosequence. This is especially true for the epigeal taxa with 78–94% carnivores in all zones. Although they do not require plants for food, the carnivores may nevertheless make use of vegetation for shelter. Furthermore, the substantial proportion of pioneer herbivores in Zone 1 (24%), which includes both phytophagous and mycetophagous taxa, decline through Zones 2–4 as the vegetation develops, before rising again in Zones 5 and 6.
Although persistence of taxa is high throughout the succession with 86% of the taxa and 79% of the epigeal taxa from Zone 1 also occurring in Zone 6, by Zone 4 the number of last appearances approximates the number of first appearances (Figure 2b), suggesting replacement change, as does the slow rise in the proportion of herbivores in Zone 5. However, there is no evidence of the species populations being affected by replacement change to the extent reported by, for example, Kaufmann (2001) and Kaufmann and Raffl (2002) who demonstrated rapid taxa turnover and almost no taxa overlap between the youngest and oldest zones at the Rotmoosferner, Austrian Alps. This may relate to greater complexity of communities and less severe environmental conditions at more southerly latitudes.
Differences in the invertebrate populations at the glacier–foreland boundary between late-successional Zone 5 and mature Zone 6 (which represents a potential end point to succession or ‘climax’ stage) imply that succession on the oldest parts of the glacier foreland has not ended. Most pronounced of these differences is the much larger number of taxa (32 taxa in Zone 6, in contrast to 5 in Zone 5) that attain peak abundance outside the foreland (Figure 2c). Another difference is the greater number of taxa (55 taxa outside the boundary, in contrast to 39 inside), although the two zones have a majority of taxa in common (Table 2) and there is little difference in terms of the number of epigeal taxa or functional groups.
The classification and ordination analyses shed further light on changes in invertebrate community structure. Although particular taxa comprising the TWINSPAN species groups exhibit individuality in their distributions, as exemplified in Figure 3, successional status is readily interpretable as the groups occupy distinctive, though overlapping positions in the chronosequence. The TWINSPAN site groups are even more age-dependent as the sites within each group generally occupy spatially contiguous areas of the chronosequence, which do not overlap with the areas occupied by other groups (Figure 4). Similarly, the disposition of site groups in the NMS biplot (Figure 5) is interpreted as a rapid successional development (indicated by the arrow) of several recognizable but indistinct communities.
Subalpine versus alpine succession
The subalpine chronosequence at Fåbergstølsbreen has similarities and differences in comparison with the chronosequence previously reported by Vater (2012) from Storbreen in the low-alpine belt of the alpine zone. Broadly similar numbers of taxa occur on the foreland but trapping rates tend to be higher at Fåbergstølsbreen where there is a similar trend of increasing numbers of taxa with terrain age reflecting similar numbers of first appearances and last appearances. Most early colonizers are therefore persisting into relatively old zones on both forelands where the mature birchwood tends to be richer in species than the mature alpine heathland. In both chronosequences, relatively few taxa are limited to terrain of intermediate age and therefore distinctly mid-successional in status. However, in the case of the subalpine chronosequence species groups 5 and 6 are characterized by small numbers of mid-successional species. In terms of functional groups, both chronosequences are dominated by carnivores, even though herbivores are more important in the subalpine zone, both in the pioneer and mature/ ‘climax’ stages. Although they are relatively few in number, omnivores also appear more important in the subalpine zone, especially in the mature stages.
Differences between the two chronosequences are clearer in relation to taxa abundance, particularly with respect to the zones in which peak catches occur, which reflect more rapid succession in the subalpine zone. This is particularly apparent in relation to Mitopus morio, the dominant pioneer species on young terrain on both forelands, which is three times as abundant in Zone 1 (deglacierized for <20 years) at Fåbergstølsbreen, and reaches its peak abundance in Zone 4 at Storbreen. A large number of other taxa exhibit a similar pattern, reaching peak abundance in Zones 1 or 2 at Fåbergstølsbreen but Zone 4 at Storbreen. This rapid expansion of the subalpine populations is followed by an earlier decline (Zone 3 at Fåbergstølsbreen; Zone 5 at Storbreen), which appears to relate to the earlier onset and greater extent of competition from later colonizers, such as Formica lugubris (which is absent from the alpine zone). The fact that many more taxa attain peak abundance in Zone 6 in the subalpine zone may reflect the availability of a greater range of available niches in the woodland.
The subalpine chronosequence is also characterized by the development of a better organized and more complex community structure than the alpine chronosequence. There are clearer differences between species groups and especially between site groups, albeit with considerable continuity reflecting individuality in the distribution patterns and behaviour of the taxa. The greater importance of herbivores and omnivores in the later stages of the subalpine chronosequence suggests the development of more complex food webs there. Furthermore, each site group occurs in fewer age zones in the subalpine chronosequence (Table 3), and the site classifications suggest fewer (three rather than six) interpretable communities in the early stages of succession at Fåbergstølsbreen, which appear to be converging on one or two mature communities. There may also be greater community differences across the glacier foreland boundary at Fåbergstølsbreen, resulting from both the greater number of rare species and the greater abundance of some of the commoner taxa in Zone 6. However, a discontinuity is not evident in Figure 5, which emphasizes that the late-successional and mature communities have much in common, and hence that succession on the glacier foreland is arguably approaching ‘climax’ status.
Implications for the ‘addition and persistence’ model
In essence, the ‘addition and persistence’ model of invertebrate succession proposes a process whereby, through time, taxa colonize newly deglacierized terrain without replacement change of discrete communities (Vater, 2012). Several aspects of the results from Fåbergstølsbreen are in accord with this model: the number of taxa caught increases across the chronosequence while the number of first appearances of taxa exceeds the number of last appearances; and most taxa recorded in the youngest zone also occur in the oldest zone, with far more taxa attaining peak abundance in the oldest zone than in any other. Thus, few taxa are exclusive to particular communities.
Nevertheless, the abundance patterns of individual taxa and the community differences do provide evidence consistent with an element of replacement change. In particular, the dominant pioneer species early in succession, Mitopus morio, can be interpreted as being replaced by Formica lugubris by Zone 5. Trombidium holosericeum and Amara quenseli, which exhibit similar distribution patterns, may also be locally dominant pioneers. Similarly, Pardosa lugubris and other species of Pardosa may be locally dominant in the oldest zones. As relatively few other taxa exhibit the same patterns, however, this cannot be said to constitute replacement of one community by another. Although a relatively large number of taxa appear for the first time in Zone 6, their low abundance means they are rarely consistent members of the late-successional or mature communities. These late colonizers are therefore additions to, rather than replacements of, those taxa that persist from the earlier successional stages.
Similarly, although there are differences between the communities, analyses at the community level show that the defined TWINSPAN groups are far from discrete entities (see, especially, Figure 5). The overall picture is one of rapid colonization producing a continuum of variability, which is organized along a successional time gradient. This does not imply an absence of community structure, only that the taxa exhibit a level of individualistic behaviour that is too high to constitute discrete communities.
It might be argued that inclusion of the flying taxa is biasing the sample in favour of the ‘addition and persistence model’ because (1) they are able to move relatively freely across the chronosequence and (2) most have not been identified to species level. Thus, they could be regarded as vagrants that do not belong to the local community characteristic of particular terrain ages, and differences in distribution and successional status between related species may be hidden within the higher taxonomic units. However, the greater mobility of flying invertebrates does not mean they should be excluded, a priori, as part of the community, especially as the abundance of many of these taxa indicates that populations have established locally. The temporal patterns identified from considering all trapped taxa are, moreover, strongly supported by those inferred from separate analysis of the epigeal taxa, most of which have been identified to species level. In relation to the pioneer taxa from Zone 1, for example, 86% of the pioneer taxa persist to Zone 6, while 79% of the epigeal pioneers persist. Furthermore, if only epigeal taxa identified to species level are considered, 80% are still classifiable as persistent pioneers.
As there is even less evidence for replacement change in the results from Storbreen in the alpine zone, it can be suggested that the ‘addition and persistence’ model is most appropriate in severe environments, where succession is a slower process and the early successional stages last longer. This conclusion is consistent with other results, not only from other Norwegian alpine and polar chronosequences (Bråten et al., 2012; Hågvar, 2010, 2012; Hågvar et al., 2009; Hodkinson et al., 2001, 2002, 2004) but also from the Austrian Alps, where replacement change has been widely recognized (Janetschek, 1949; Kaufmann, 2001; Kaufmann and Raffl, 2002). However, the model needs to be tested further under a wider range of environmental conditions, in relation to particular groups of invertebrates, and in the context of secondary succession.
Conclusions
Individual invertebrate taxa colonize the glacier foreland rapidly and a high proportion persist from the pioneer zone to the mature communities. Of the 67 recorded taxa, almost 50% occur on terrain within 20 years of deglacierization, and 86% of these pioneer colonizers occur also on the oldest terrain (deglacierized c. 9700 years ago). Similar patterns of rapid colonization and persistence apply to the pioneer epigeal taxa, which exhibit ~80% persistence to the mature stage.
Carnivores are particularly abundant, comprising 73–94% of epigeal taxa throughout the chronosequence. The majority of the early colonizers, including the dominant pioneer, Mitopus morio, attain peak abundance within 40 years of deglacierization. However, the flying pioneers tend to attain peak abundance sooner than the epigeal pioneer taxa, and many pioneers and most of the later colonizers attain peak abundance in the oldest zone. Later stages of succession are dominated by Formica lugubris, and are generally richer in species because of the persistence of pioneers combined with a relatively large number of additional taxa of low abundance.
At the community level, TWINSPAN species groups are related to terrain age but overlap in composition owing to considerable within-group variability reflecting taxa individuality. However, TWINSPAN site groups are more strongly related to terrain age, forming contiguous, nearly non-overlapping groups (in terms of presence and abundance) across the chronosequence and in the NMS ordination. Thus, the invertebrate communities are not discrete entities but are interpreted as parts of a continuum of variability. They nevertheless represent the emergence of a community structure that appears better organized than on glacier forelands in the alpine zone.
Invertebrate succession in the subalpine zone is therefore consistent with a modified ‘addition and persistence model’. This allows for an element of replacement change in the later stages of a successional process that is nevertheless largely characterized by the addition of new taxa and the subsequent build-up of their individualistic populations. There appear to be few dispersal limitations on the colonization of a large proportion of the epigeal taxa (as well as the flying taxa), which reproduce successfully soon after arrival on the glacier foreland. Later changes in the abundance of invertebrate taxa and community organization during succession can be attributed to two processes: first, competition between predatory carnivores (providing the main underlying mechanism of change that is relatively independent of the parallel process of vegetation succession); and second, the development of food webs involving herbivores and, to a lesser extent, omnivores.
Footnotes
Acknowledgements
We are grateful to staff at the National Museum of Wales, Cardiff, for assistance with identification; to Mauro Gobbi, Sigmund Hågvar and Lawrence Walker for their critical and constructive comments on the manuscript; and to Anna Ratcliffe for drawing up the figures for publication.
Funding
This paper constitutes Jotunheimen Research Expeditions Contribution No. 184. Fieldwork was carried out on the University of Wales Swansea, Jotunheimen Research Expeditions, 2003 and 2004. Funding was provided by the Geography Department, School of the Environment and Society, University of Wales Swansea and the Jotunheimen Research Trust.