STUDY AREA

Kasatochi Island occurs in the “Aleutian Islands Ecoregion,” characterized as a naturally treeless region where dwarf shrub communities predominate at higher elevations and on sites exposed to the wind, and where herbaceous communities occur on more protected sites (Gallant et al., 1995). Species comprising the vegetation of the Aleutian Islands are characteristically perennial. Küchler (1966) classified the former dwarf shrub communities as “Aleutian heath and barren (Empetrum-Vaccinium),” and the latter herbaceous communities as “Aleutian meadows (Calamagrostis-Anemone).” A vegetation map of Kasatochi Island (Fig. 1) shows five map units; these are listed in descending order of areal extent: (A) “short grass meadow” community, occurring on lower and middle slopes; (B) “bare rock” around the crater rim and on the marine shoreline; (C) “vegetated talus” on lower and middle slopes ; (D) “lichen/moss community” at upper elevations adjacent to the crater rim; and (E) “dense Leymus” at lower elevations with some isolated patches on middle slopes.

Figure 1

Pre-eruption vegetation of Kasatochi Island, Alaska, in 1996 (modified slightly from Scharf et al., 1996). Numbers on the map indicate relevé locations; these relevés were sampled during three periods: the pre-eruption periods 13 August 2001 (relevés 1–4) and 20 August 2003 (relevés 5–7), and the post-eruption period from 10 to 12 August 2009 (relevés 8–21).

Kasatochi Island is an emergent summit of a predominantly submarine volcano that last erupted in 1828 (Miller et al., 1998). Prior to the 2008 eruption, the island consisted of a single, undissected cone with a central lake-filled crater about 0.75 km in diameter and steep flanks (18–45°) rising to a maximum elevation of 314 m. The pre-eruption soils of Kasatochi were mapped as rough mountainous land composed of steep rocky slopes with thin soils occurring on lower slopes and valleys (Rieger et al., 1979); data from surrounding islands suggest that Typic Cryandepts were likely well-represented on Kasatochi. A characterization of the pyroclastic deposits and pre-eruptive soil following the 2008 eruption of Kasatochi Island is given by Wang et al. (2010 [this issue]). Pollen studies of Aleutian soil profiles show a repeated pattern of volcanic ashfalls followed by vegetation recovery (Anderson and Bank, 1952; Heusser, 1978).

Climatically, Kasatochi Island is within the northern boreal subzone described by Tuhkanen (1984) in the hyperoceanic (O₂) sector and humid (h) province. Based on available records (1971–2000) from Adak Island, the nearest climatic station (51.88°N, 176.65°W), 80 km southwest of Kasatochi, the mean annual temperature and precipitation recorded are 4.6 °C and 1372 mm, respectively (WRCC, 2009); August is the warmest month with a mean temperature of 10.5 °C (Fig. 2). Multivariate analysis of climate along the southern coast of Alaska indicates that the Aleutian Islands form a distinct climatic group with moderately cool temperatures in the fall, winter, and spring and much colder summers than other coastal stations (Farr and Hard, 1987).

Figure 2

Climate diagram for Adak Island, Alaska. Abscissa shows the months January to December. Ordinate: the first dash-mark upwards represents 10 °C and 20 mm, respectively. Name of station given in upper left-hand corner with latitude/longitude and elevation given below. Number of years data were collected for temperature and precipitation given in brackets, followed by mean annual temperature and mean annual precipitation. Top curve is mean monthly precipitation and lower curve is mean monthly temperature. The diagram was prepared following Lieth et al. (1999).

VEGETATION AND ENVIRONMENTAL SAMPLING

We sampled vegetation in twenty-one 25 m² relevés (Westhoff and van der Maarel, 1973) during three periods: the pre-eruption periods 13 August 2001 (relevés 1–4) and 20 August 2003 (relevés 5–7), and the post-eruption period from 10 to 12 August 2009 (relevés 8–21). The pre-eruption relevé sites were selected to represent a spectrum of structural and compositional variation in the landscape from lower to upper mountain sites, but excluded beaches. The post-eruption sites were selected by searching for plants from beaches to within 100 m of the crater rim and recording a relevé wherever plants occurred. Safety concerns precluded sampling near the post-eruption rim. Cover abundance was estimated for all vascular plants, bryophytes, and macrolichens according to the nine-point ordinal scale of Westhoff and van der Maarel (1973). Although not part of our study, fallout collectors for sampling insects (Sikes and Slowik, 2010 [this issue]) were utilized to collect samples for wind-dispersed plants (Edwards and Sugg, 2005). Ten 0.1 m² collectors were placed 10 m apart along a 100 m transect oriented from NW to SE on the westernmost coast of the island on 14 June 2009 and recovered on 10 August 2009. Plant nomenclature follows USDA, NRCS (2009). Voucher specimens were deposited at several herbaria: University of Alaska Museum (ALA), fungus; State University of New York, Binghamton (BING), Sphagnum; Canadian Museum of Nature (CAN), Poaceae and Salix; Iowa State University (ISC), Botrychium; Missouri Botanical Garden (MO), Epilobium; New York State Museum (NYS), Platanthera; University of Waterloo (WAT), Stellaria and Cerastium; University of British Columbia (UBC), bryophytes and lichens (2003 vouchers); U.S. National Herbarium (US), Poa; and University of Wisconsin (WIS), lichens (2001 vouchers).

We characterized the local environment at each relevé by recording aspect (°), elevation (m), litter cover (%), slope inclination (°), latitude and longitude (with a GPS using WGS84 datum). Three to five samples of the top 15–20 cm of post-eruption material (henceforth called soil) from the center of each relevé were collected, pooled, and placed in plastic zip-lock bags; these were kept cool until analysis. Where present (n  =  7 relevés), pre-eruption soil horizons were also sampled in the same manner, but up to 30 cm depth. Soils were not collected during the 2001 and 2003 season, although soil texture was estimated (Burt, 2004).

VEGETATION CLASSIFICATION

Our data set of 21 relevés and 91 species was entered in TURBOVEG (Hennekens and Schaminée 2001). We used the method OPTIMCLASS in JUICE 6.5 (Tichý, 2002; Tichý et al., 2010) to determine the optimal number of relevé groups. Numerical analysis was performed with the classification and ordination methods of the MULVA-5 program (Wildi 1989; Wildi and Orlóci, 1996), following Talbot et al. (2010). In brief, relevés were first clustered into relevé groups using the similarity ratio and complete linkage, and species were clustered into species groups using a similar procedure. Next, analysis of concentration (Feoli and Orlóci, 1979) ordered dense species-relevé groups along the diagonal (the main floristic gradient) according to the first axis in correspondence analysis (Hill, 1974). A set of highly discriminating species was then determined using Jancey's (1979) ranking.

An ordination was performed using CANOCO 4.5 (ter Braak and Šmilauer, 2002) with the WinKyst1.0 Add-On (Šmilauer, 2003) to provide non-metric multidimensional scaling (NMDS). The species data were transformed using a square-root transformation and a distance matrix calculated using Bray-Curtis distance. The file of 91 species was treated in CANOCO 4.5 and CANODRAW 4.1 as suggested by Šmilauer (2003) with environmental data related to the sample scores in an unconstrained analysis.

SOIL ANALYSIS

Soils were analyzed by the Kuo Testing Labs, Inc. (KTL), Othello, Washington. All methods followed Burt (2004). Organic matter was analyzed by the Walkley-Black method; orthophosphate by Olsen's extraction; cations by inductively coupled plasma spectrophotometry; nitrate-N and ammonium-N by colorimetric analysis of a KCl extract; SO₄-S, Boron, Zn, Mn, Cu, and Fe by extraction with a DPTA-Sorbitol solution; texture by the hydrometer method; and organic matter by ashing in a muffle furnace.

GENERAL FEATURES

Typically, the plants we found on post-eruption Kasatochi during 2009 were relatively isolated and localized, and particularly occurred on lower mountain slopes and beaches; no plants were found growing on middle and upper mountain slopes. The soils from the 2008 eruption were undergoing rapid erosion, and occurred on upper mountain slopes (rills to 30 cm deep), increasing in depth on lower slopes (gullies to several meters deep). Such active erosion resulted in low substrate stability. Plant diversity was highest in the southern and southeastern corner of the island and was associated with the presence of antecedent soil horizons particularly at the top of a bluff ridge and on lower mountain slopes where rill erosion exposed the pre-eruption soil.

The number of taxa recorded in the relevés totaled 91 (51 vascular plants, 23 lichens, 16 bryophytes, and 1 fungus; Table 1Table 1); of these, 86 species were recorded in the pre-eruption landscape, 18 in the post-eruption, 13 occurred in both the pre- and post-eruption, and 5 were only recorded in the post-eruption landscape.

Table 1

Vegetation of Kasatochi Island, Alaska, including pre-eruption type (RG2) and post-eruption types (RG1, RG3, RG4, RG5). Key to types: RG1—Festuca rubra cliff shelf; RG2—Lupinus nootkatensis–Festuca rubra meadow; RG3—Honckenya peploides beach meadow; RG4—Leymus mollis bluff ridge (and beach); RG5—Aleuria aurantia lower slope barrens. The order of relevé and species groups is based on analysis of concentration (Feoli and Orlóci, 1996) along the main floristic gradient according to the first axis in correspondence analysis (Hill, 1974). Table 1 continues on next page.

Table 1 (cont.)

Species of low presence and diagnostic value usually occurring three times or less. Each species name is followed by relevé group number (RG), releve number, and cover-abundance value.

CLASSIFICATION AND ENVIRONMENTAL CHARACTERISTICS OF VEGETATION TYPES

Numerical-phytosociological analysis of all relevés (pre- and post-eruption) identified 5 vegetation or community types, hereafter called relevé groups (RGs), and 8 species groups. The 5 (relevé groups) × 8 (species groups) contingency table, ordered in Table 1 along the main floristic gradient ( =  the ordering of the species and relevés along the first axis of the correspondence analysis), showed block structure that deviated from random expectation. A mean square contingency coefficient of 0.442 indicated that group structure was relatively strong (Wildi and Orlóci, 1996). A quantitative display of relationships among the 5 RGs is presented in Figure 3.

Figure 3

Classification dendrogram of five relevé groups (community types) from Kasatochi Island, Alaska, obtained using the van der Maarel coefficient and complete linkage (farthest neighbor) clustering (Wildi and Orloci, 1996) of 21 relevés. The dendrogram was prepared using ClustanGraphics7 (Wishart, 2003).

The 5 RGs showed differences in species composition and dominance (Table 1), vegetation structure (Table 2), and site and soil characteristics (Tables 2 and 3). A summary of each type follows (ordered as in Table 1, along the floristic gradient). Percent cover, elevation (m), and slope (°) are expressed as mean values:

Table 2

Vegetation structure and environmental variables of five community types from Kasatochi Island, Alaska.

Table 3

Ecological data for four relevé groups of Kasatochi Island, Alaska, vegetation. No soils were collected from RG2. Subgroups: A  =  post-eruption soil material from center of the relevés; B  =  pooled pre-eruption soil material from within relevés.

ORDINATION OF RELEVÉS

Nonmetric multidimensional scaling ordination (Fig. 4) showed 21 relevés in 5 relevé groups. The primary NMDS solution for two dimensions was found with 41 iterations out of 100, stress  =  0.12133 indicating a good fit. The eigenvalues of the first two axes of PCA, applied to NMDS configuration to rotate it so that the spread of relevés was maximized along the two axes, showed that the resulting first axis had pronounced explanatory value, compared with the second one (0.663 vs. 0.337).

Figure 4

Non-metric multidimensional scaling ordination (Šmilauer, 2003; ter Braak and Šmilauer, 2002) of 21 relevés with projected environmental variables from Kasatochi Island, Alaska. The scatterplot shows the five cluster models of community types identified in classification dendrogram (Fig. 3). Solid symbols distinguish the pre-eruption group (RG2) and hollow symbols distinguish post-eruption groups (RG1, RG3, RG4, and RG5). Key to types: RG1—Festuca rubra cliff shelf; RG2—Lupinus nootkatensis–Festuca rubra meadow; RG3—Honckenya peploides beach meadow; RG4—Leymus mollis bluff ridge (and beach); RG5—Aleuria aurantia lower slope barrens.

The 5 relevé groups were relatively distinct on the ordination biplot (Fig. 4) and were delineated by slope and elevation. The pre-eruption community (RG2) occupied a central position in the ordination and most closely resembled one relevé in post-eruption community RG4 (relevé no. 14) because of several shared species.

SPECIES OCCURRING OUTSIDE THE RELEVÉS

Several species were found that were not sampled in the relevés, including two crustose lichens, two mosses, and two vascular plants. The lichens included Caloplaca sp. and Verrucaria sp. growing on rock outcrops in the salt-spray zone. The mosses were Eurhynchium praelongum, found growing around a small volcanic vent, and Funaria hygrometrica that included old capsules and new gametophytes (O. Lee, UBC Herbarium, personal observation). The vascular species included: a fern, Cystopteris fragilis, found in a rock crevice; Ligusticum scoticum, a plant typical of seashores that occurred on a bluff ridge; and Senecio pseudoarnica, a species characteristic of Aleutian beaches, which was associated with antecedent soil (J. Williams, personal observation).

COMPARISON WITH OTHER BOREAL REGIONS

The long-term knowledge gained from the study of the eruption of the Icelandic island of Surtsey provides useful comparative data (Jakobsson et al., 2009). Kasatochi and Surtsey differ in their origins; Kasatochi is of Late Tertiary and Quaternary origin, whereas Surtsey first emerged in 1963. Nevertheless, there are strong similarities in climate and flora. For example, the mean annual temperature and precipitation of Surtsey is 4.8 °C and 1589 mm (Fridriksson, 2005), respectively (Fig. 2), while corresponding values on Adak are 4.6 °C and 1372 mm, respectively. The floristic similarity between the Surtsey and the Aleutian Islands is high with 58% (36 species) of the total vascular plant flora of Surtsey (62 species; Fridriksson, 2005) occurring in the Aleutian Islands (cf. Hultén, 1968). Two of the first pioneer species on Surtsey, Honkenya peploides and Leymus mollis, were common pre- and post-eruption species on Kasatochi Island. Throughout the Aleutians these species are abundant and form the basis of the main community type at the leading edge of the vegetation along the shoreline (Viereck et al., 1992; Talbot and Talbot, 1994; Talbot et al., 2010).

Although not closely monitored over time, Bogoslof Island in the eastern Aleutian Islands may serve as a useful baseline for the speed of vegetation origin, via colonization (but not survival of pre-eruption plants), in the Aleutian Islands, and subsequent succession. Byrd et al. (1980) summarized the changes in the island's flora, which are abstracted here. On Bogoslof, which emerged as a new island due to volcanic activity, the vegetation began to appear in 1817, twenty-one years after its origin in 1796. Over many years, periodic reports of the vegetation alternated between records that vegetation was present or that it was absent, but in 1916, patches of vegetation (primarily “coarse grass”) began to take hold, and by 1946 the island's plateau was covered with “tundra vegetation.” By 1973, the plateau was completely vegetated. Volcanic activity continued periodically in the vicinity of Bogoslof with the most recent eruption in 1992. A total of 12 vascular plants are reported (Byrd et al. 1980), notably Honckenya peploides, Leymus mollis, Ranunculus occidentalis, and Senecio pseudoarnica, which are on post-eruption Kasatochi. Colonization and establishment of mature plant communities typical of other naturally treeless Aleutian islands appear to have proceeded quite rapidly on Bogoslof, over ca. 180 years. This is quite different from other boreal areas; Grishin et al. (1996) estimated vegetation recovery on Ksudach Volcano in Kamchatka will take more than 2000 years, where the climate supports forest vegetation on zonal sites. It must be reiterated that there are differences between these areas in terms of climate and plant communities. On Kamchatka, forests of Betula ermanii reach their mature stage (not the final stage of succession) at an age of 120–150 years in favorable conditions with good soils. In addition, soil formation on Kamchatka occurred on unstable, acidic (pumice) substrates over 1 m thick in climatic conditions with temperature minimums below −20 °C (Grishin et al., 1996, 2000). Accordingly, mature plant communities on Bogoslof are not quite the same as “final stage of succession,” a term equal to climatic climax that Grishin et al. (1996) meant in their study for estimating the term of succession.

The coastal plant communities of the Aleutians are typically disturbed by surf and wind erosion and their species selected for quick successional response, and observations from Bogoslof suggest that colonization, in addition to recovery via extant vegetation, will likely play a role in the revegetation and succession on Kasatochi as on Bogoslof.

In post-eruption colonization studies of the boreal zone—Hokkaido (Tsuyuzaki, 2009), Kamchatka (Dirksen and Dirksen, 2007), and southwestern Alaska (Griggs, 1933)—several factors were common to all in the reconstruction of natural communities. These included spatial and temporal heterogeneity (particularly ash depth) that played a major role in influencing whether preexisting plants were able to survive an eruption. Generally, the greater the distance from a volcano, the greater the decrease in the depth of the deposit; this allowed plants to come up through deposits less than about 30 cm deep. This factor, combined with the existence of buried soils with surviving underground vegetative structures such as rhizomes, permitted local survival; the plants persisted underground. Rain erosion typically cut down through the ash, exposing the antecedent soil to sunlight and warmer temperatures and facilitating recovery of buried plants. These buried soils provide a less hostile substrate for plants than ash (in sense of stability of ecological regimes). Antecedent soils are important for providing a substrate both for surviving plants, and for the roots of seedlings or spores that have been deposited and germinate on thin layers of surface ash. The presence of refugial microhabitats also played an important role in a legacy effect in all these cases, permitting plants to spread from survival foci. The lee offered by obstacles sheltered habitats or organisms from the effects of ash fall or pyroclastic flows. Each of these factors applies as well to Kasatochi Island.

ORIGIN OF THE POST-ERUPTION FLORA ON KASATOCHI

Plant propagules are expected to reach Kasatochi using different dispersal mechanisms. Cody (2006) proposed three major means of interisland transportation of propagules: anemochory, zoochory, and hydrochory. Anemochores travel on the wind. Examples of post-eruption species found on Kasatochi that might use anemochory include plants with plumed seeds such as Achillea millefolium var. borealis, Anaphalis margaritacea, Senecio pseudoarnica, and Epilobium hornemannii ssp. behringianum, and the spores of Cystoperis fragilis and Ceratodon purpureus. Fern and moss spores are light—many have longevity ranging from several weeks to over 100 years—and thus have great dispersal potential by wind (Cody, 2006), as do bryophyte diaspores (Schofield, 1985). However, the Cystopteris fragilis observed in the protected rock crevice is unlikely to have derived from spores, given there has been no time to pass through the gametophyte stage. The vegetative parts of mosses, although heavier than spores, can also be wind dispersed (Fridriksson, 2005). Marchantia polymorpha is widespread and common in the Aleutian Islands (Davison, 1993). Young and Klay (1971) provided evidence of its apparent long-distance dispersal to a crater of Deception Island, Antarctica, and it may have arrived on Kasatochi via the same mechanism. Alternatively, the spores of Marchantia may have survived the eruption in a soil bank. Thus, some doubt exists as to its presence either from long-distance dispersal or spore bank.

Zoochores are distributed by animals such as gulls that spread seeds in the process of gathering nest material on Bogoslof (Morris, 1937) or snow buntings that distributed seeds on Surtsey (Fridriksson and Sigurdsson, 1968). Notably, we discovered five gull nests with nest material made from plants. The nest material was typically dominated by dead vascular plant litter and brown algae but one nest contained a living culm of Leymus mollis. The Leymus specimen had not taken root but in the future, plant matter in nests may account for some colonization, as it did on Bogoslof (Morris, 1937; Gabrielson and Lincoln, 1959).

Hydrochores are spread by sea currents; Honckenya peploides and Leymus mollis are good examples. Cody (2006) pointed out that Honckenya has dehiscent capsules and brittle stems that can break easily by wave action; they are often found washed up on beaches throughout the Aleutians, including Bogoslof. We consider hydrochory unlikely as the time period required for establishment via drifting was short, and our reconnaissance of the beaches both in June and August 2009 revealed no vascular plant litter. Furthermore, Kasatochi Island continued to increase in size from the erosion of upper slope ash. To become established any floating plant would have had to overcome this instability. Individual plants of these two species probably traveled downslope and became established on back beaches some distance from the sea.

On what appeared to be an island smothered in volcanic ash—in some places meters thick, where continuation of plant life seemed improbable—living plants were observed as mature, but mostly vegetative, perennial plants, and seedlings. We consider that the majority of plants observed in post-eruption sampling were survivors of the Kasatochi eruption, rather than colonizers. One relevé (no. 14, RG4) contained an assemblage of several mature plants suggesting a species-poor relict of the previously more widespread pre-eruption RG2 type; this assemblage and similar ones not yet discovered could contribute to legacy effects in the revegetation of Kasatochi.

A number of factors contributed to the survival of at least a fragment of the original vegetation as reported previously for other boreal regions (see above). These included eruption heterogeneity such as thin ash deposits on bluff ridges that exposed the antecedent soils. Similarly, rill erosion cut through the ash overburden and exposed the antecedent soil. Volcanic ash can be removed by rain erosion quite rapidly on volcanic slopes (Thornton, 2007). The depth of the rill erosion was an important factor in exposing the antecedent soil but not eroding it away. Some of the pre-eruption soil was eroded; seeds or vegetative propagules were then likely carried downslope by gravity and intermittent streamlets. Given the short growing season in the Aleutians, the size of the plants that developed indicated that they arose from a vegetative portion of plant buried in the antecedent soil.

Seedlings were observed intermixed with mature plants both in June, and in August 2009. We suggest that seedlings arose mostly from seed beds, because (1) no mature plants observed in June or August had seeds; (2) no seeds were collected in insect traps set out on post-eruption Kasatochi, suggesting an absence of diaspores in the early post-eruption environment.; (3) the distance between islands and the short time period elapsing since the eruption suggest colonization via hydrochore seeds is unlikely; (4) no seedlings were found along the beaches, only mature plants; and (5) uncharred organic material in the soil suggested seeds could have survived. Seedlings derived from zoochory were unlikely, as terrestrial avian species typically associated with seed dispersal elsewhere such as Surtsey (Williams et al., 2010 [this issue]), were absent.

The post-eruption landscape provided a number of microhabitats conducive to the survival of plants. The presence of old moss capsules of Funaria hygrometrica in the post-eruption environment suggested that mosses survived the eruption intact. Cliff shelves were protected by overhanging portions of the cliff walls from repeated pyroclastic flows that emanated from the volcano caldera and cascaded downslope. These cliff shelves served as refugia where living, loosely caespitose clumps of intact plants survived, because they were not covered for long periods with ash. It is clear that the north and south side of the island differed with regard to provision of these microhabitats. The north side, with steeper slopes and more cliffs than the south side, provided more microhabitats.

We observed buried organic matter in antecedent soils. An extensive bluff ridge on the southern part of the island had a thin layer of ash overburden over the antecedent soils on the ridge top. The pyroclastic flows appeared to have travelled quickly over the ridge top leaving a heavier deposit of ash above and below the bluff top. Many of the plants we observed peaked in their abundance on this ridge top where the thin ash layer left the antecedent soil exposed. Our 25 m² relevés at these bluff ridges included both the above and below portions of the ridges, and some ridge plants had been transported down slope by gravity into the lower portion of the bluff.

In future monitoring of plant succession on Kasatochi Island, we will address the relative importance of survivors versus new arrivals in determining change, the dynamic interplay between actively changing topography and microsite availability for colonization, and the establishment and replacement of plant communities. To accomplish this, it will be necessary to identify the source of plants; the application of genetic data can be used for this purpose (Yang et al., 2008). We suggest growth chamber laboratory experiments be initiated on antecedent soils to determine if seedlings will develop from the seeds that might be buried in the soils. In addition, vegetation analysis of a geologically similar nearby island such as Koniuji Island (25 km east of Kasatochi) might shed further light on potential successional pathways on Kasatochi. As recommended by Thornton (2007), investigations should continue with regular and frequent visits and surveys should be done at short intervals, to identify and monitor stable legacy communities as well as new colonizers. We consider it imperative that the impact of anthropogenic activities be minimized.