HAWAI`I VOLCANOES Invasion and Recovery of Vegetation after a Volcanic Eruption in Hawaii NPS Scientific Monograph No. 5

Studies in Hawaii

A framework for ecological studies in the Hawaiian Islands has been given by several descriptions of vegetation zones (Hillebrand 1888; Rock 1913; Hosaka 1937; Egler 1939; Robyns and Lamb 1939; Hart and Neal 1940; Ripperton and Hosaka 1942; Krajina 1963; Knapp 1965). Three ecological studies were concerned with more local physiographic sections of Oahu (Hosaka 1937; Egler 1947; Hatheway 1952). Their emphasis was on the description of current plant communities. Fosberg (1961) provided a summary description of about 30 major ecosystems that are prevalent on nearly all the high islands.

Several ecological studies deal specifically with the island of Hawaii, which because of its recent volcanic surfaces, shows major unconformities in vegetation types to the other high islands. Doty and Mueller-Dombois (1966) reviewed all bioecological studies that had been done up to that time in Hawaii Volcanoes National Park. In addition, a detailed framework of ecosystem types was provided with this publication—an aerial photo vegetation map and five topographic vegetation profiles. These ecosystem types were further investigated for their phytosociological relationships (Newell 1968) and tree stand relative to soil characteristics (Rajput 1968). Mueller-Dombois (1967) analyzed in some detail the ecological relations in the alpine and subalpine vegetations on Mauna Loa volcano, and a comparison of east-flank vegetations on the still active Mauna Loa and the older, dormant Mauna Kea volcanoes was made (Mueller-Dombois and Krajina 1968). These studies provide the more specific ecological framework for studies dealing directly with succession on volcanic surfaces.

Studies directly concerned with aspects of vegetation dynamics on new volcanic material of Hawaii, the youngest island, were done by Forbes (1912), MacCaughey (1917), Robyns and Lamb (1939), Skottsberg (1941), Doty (1957, 1961, 1966, 1967a, 1967b), Miller (1960), Fosberg (1959), Mueller-Dombois (1967), Jackson (1969), Atkinson (1969, 1970), and Eggler (1971).

Forbes was the first to study plant succession on recent Hawaiian lava flows. His observations were confined to the summer-dry region on the lee side of Mauna Loa. Here he studied five lava flows with dates of 1859, 1884, 1887, 1907, and one of recent origin but not dated. His major conclusions with regard to plant invasion, succession, and climax on lava flows on the leeward side are summarized in the following points:

1. Appearance of lower cryptogams, eventually becoming conspicuous on the a'a.

2. Appearance of Polypodium pellucidum (folded form), Sadleria cyatheoides and Metrosideros polymorpha (ohia), first on pahoehoe, and at a much later date on a'a.

3. Gradual development of the typical floral aspects of the immediate vicinity, if in the central region of an ohia forest.

4. Establishment of the final native vegetation, if in the central region of a koa forest.

5. A later stage may be the encroachment of the naturalized flora, due to a change of conditions brought about by human agency.

MacCaughey (1917) also confined his study to the xerophytic regions. He reported that the rate of invasion depended on rainfall and adjacent vegetation, and that his findings generally agreed with those of Forbes. In contrast he found that lichens occurred much sooner on a'a than on pahoehoe, while ferns and trees established earlier on the pahoehoe lava type.

Robyns and Lamb (1939) recognized and classified five major climax formations for Hawaii. In so doing they followed the monoclimax concept of Clements. They emphasized that climate controls the final form of vegetation, while the soil accounts only for developmental stages. They concluded that the rate of invasion and vegetation density increased with rainfall and that moisture is more important in plant establishment than age of the substrate. They reported the course of primary succession in the Kilauea region in three steps:

1. Invasion of cracks in the new flow by ferns and flowering plants that are common to the adjacent area, and supported by nonvascular cryptogams especially on the a'a lava.

2. Gradual building up of heavier plant covering, filling in between the cracks, producing a shrub stage in which Dodonaea viscosa, Styphelia tameiameiae, and Metrosideros polymorpha predominate.

3. Development of a plant community typical of the vegetation formation found in the surrounding area.

Skottsberg (1941) was the first investigator in Hawaii to make observations in the same locations after a lapse of time by using permanent quadrats to study plant succession. He established six 10 x 10-m quadrats in 1926. Four were laid out on the 1920 Kilauea flow in the Kau Desert in a summer-drought climate, the other two were established on the 1919 Mauna Loa flow in a rain-forest climate in southwest Hawaii. Within each of the flows, he compared a'a with pahoehoe sections. As expected, the plant invasion rate was very much faster on the 1919 flow (wet climate) than on the 1920 flow (dry climate), and the species diversity was greater in the moist climate. In general, he found invasion of vascular plants to be denser on pahoehoe, while cryptogams (mosses and lichens) were becoming widely established on the a'a lava. This supported the earlier observational studies. Skottsberg's work points out the need of distinguishing finer substrate differences in relating plant invasion to the type of volcanic surface material.

Doty (1957, 1961, 1967a, 1967b) made the most intensive study of plant succession on Hawaiian volcanic materials to date. His work was confined to the 1955 lava flow on the east flank of Kilauea Volcano, which is in a humid climate. He observed that blue-green algae were established 3 months after the flow had stopped. At 6 months the algae were followed by a large number of individuals of cryptogamic and vascular plants. These consisted of the following species: a fungus (not named); a tree, Metrosideros polymorpha; an herb, Erechtites valerianaefolia; an orchid, Spathoglottis plicata; a fern, Nephrolepis exaltata; and a moss, Campylopus. Doty reasoned that the vascular species were established because of abundant water from steam condensing (recycling rainwater) on the flow surface. About 14 months after the eruption, large numbers of plants died. Doty assumed that a drought had developed on the flow. He believed this to be caused by a decline of vapor steaming. He attributed the reduction of steam condensation to the cooling of the flow-interior. During the drought, nonnative species disappeared completely and the number of native species was greatly reduced also. However, when rainy periods prevailed, native cryptogamic populations were found to form a succession. The blue-green alga Scytonema and then Stigonema were replaced by Stereocaulon lichens.

Fosberg (1959), after examining the alpine and subalpine zone of Mauna Loa, believed that the scattered, high altitude plants seem to have no ecological relation or dependence between them. He found that they occur in separate niches or isolated examples of the same niche. Contrary to Fosberg's findings, Mueller-Dombois (1967) found species successional relationships in the alpine scrub vegetation on Mauna Loa. He reports that where the shrubs Vaccinium peleanum and Styphelia douglasii grow together, the latter tends to replace the former.

Jackson (1969) studied the role of Stereocaulon lichen in rock weathering at low altitudes on eight dated Mauna Loa and two Kilauea flows that traverse different rainfall zones. He found that this lichen greatly accelerated the weathering of the basalt lava. A reddish-brown colloidal "gel" was isolated that consistently occurred in association with Stereocaulon on the lava flows in areas of higher rainfall. The gel was not found on rock surfaces not occupied by this lichen nor on those in dry climates, where Stereocaulon remained stunted or immature. The gel was identified as a polymorph of ferric (III) oxide containing minor amounts of aluminum and titanium oxides and traces of silicon. According to Jackson, this is the first record of this specific iron compound ever found in nature, and he tentatively gave it the name "lichenite."

Atkinson (1969, 1970) examined the successional trends in the humid and subhumid lowland to montane areas of Mauna Loa and Kilauea volcanoes from a side-by-side comparison of flows of different ages. He was trying to find a method to extend the dating of lava flows beyond the recorded dates, which go back about 200 years. By using the rate of calcium loss, pH change, and titanium gain, he believed that he had found a means of extending the lava flow dates in the humid and subhumid lowland to the montane zone for another 200 years. On this basis, he described four chronosequences as being distinct, all beginning from volcanic rock. According to Atkinson, a coastal sequence terminates via a Metrosideros stage in a Pandanus tectorius forest; a humid upland sequence, in a Metrosidero-Cibotium forest; a humid lowland series, in a Metrosideros dominated forest with minor quantities of Cibotium tree ferns. A subhumid to summer-dry series is believed to terminate in a Metrosideros-Diospyros ferrea forest. Starting from barren lava, the development of these four different forest types is thought to take place in less than 400 years. No consistent differences in succession were observed in relation to a'a and pahoehoe lavas.

Eggler (1971) studied vegetation on 16 young lava flows on the island of Hawaii for approximately 3 years. These represented a'a and pahoehoe types found in both wet and dry climates. They ranged in age from prehistoric time (prior to 1778) to deposits made in 1965. Eggler said that he could find little evidence of a succession pattern among the vegetations on these flows. Further, he observed that the rate of establishment of plants on these flows may have differed greatly. After contending that there are no distinguishable patterns in succession and rate of establishment among Hawaiian laval vegetations, Eggler proposed a formula for predicting biomass accumulations on volcanic substrates. The factors in this formula were: age of lava flow x precipitation x lava type x soil factor = biomass/ha. It seems clear that Eggler's design of side-by-side comparisons of flow vegetations did not permit reliable conclusions on succession and rate of establishment. His biomass formula fails to reveal any phytosociological relationships. These would be important for an understanding of succession.

The literature survey of plant succession studies on Hawaii shows that the information is still only fragmentary. This is due in part to the great range of environmental conditions available and in part to the differential intensity of the studies. Three general questions have been partially answered for limited types of climates and substrates. These are:

1. What kinds of plants are the invaders on new lava materials and what are their followers?

2. Which type of lava is more rapidly invaded: the rough a'a or the smooth pahoehoe?

3. How long does it take for a forest to develop from the time of lava flow origin?

No attention has as yet been paid to succession on other types of volcanic substrates, for example, the various kinds of pyroclastics. Ash substrates are as prevalent as are lava rock substrates in Hawaii, but ash or pyroclastic substrates have received no attention in succession studies. From a fundamental viewpoint, plant invasion and succession on pyroclastic materials are equally significant. Also, little is known about the rate of species increase in the initial stages of succession. Only three studies were based on periodic observations of the same places. All others relate to conclusions drawn from adjacent observations that are presumed to form chronosequences and these are still very general.

Studies in Other Volcanic Areas

A comprehensive review of work done in other volcanic areas of the world, including the Hawaiian Islands, was recently presented by Uhe (1972).

The more pertinent aspects relative to the present investigation may be summarized in three points: (1) Damage to existing vegetation and its recovery after volcanic activity; (2) New colonizers; (3) Invasion rates with respect to different materials.

1. Partial destruction of existing vegetation from deposits of the same volcanic material was found to have a selective effect on species composition and size within species. Eggler (1948, 1963) observed a better recovery of oaks than of other tree species and found that medium-sized pines were favored over small- and large-sized pines under deposits of ash on the slopes of El Paricutin, Mexico. Such effects may have contributed to differences in current species composition of vegetation islands on Mauna Loa. Eggler (1967) also found that the diameter of survival trees in the ash fallout area increased greatly. After examining the increased xylem tissue, he believed that in some cases the rapid increase in growth rate was caused by a decrease of competition. In another instance, he suggested that auxin increase, decrease, or destruction could have accounted for various rates of growth. A similar increase of the diameter of ash-buried survival trees was reported by Griggs (1922) from Mount Katmai.

Several investigators have reported that survival plants appeared to accelerate the rate of invasion of new colonizers and the revegetation process as a whole. Sands (1912) had observed great differences in the rate of vegetation reestablishment on St. Vincent Volcano (West Indies) as related to the depth of the ash blanket. Where the ash cover was shallow (implying a depth range of 30-150 cm), vegetation reestablished rather quickly. He attributed this to plant parts that survived under the ash and stated that much of the vegetation recovered from buried root systems and also from seeds. However, the latter explanation is rather questionable and would need reexamination. Gates (1914), studying the plant succession on Taal Volcano in the Philippines, also observed that where plants survived the eruption, either with their aerial parts or roots, the reestablishment of vegetation was faster. In habitats where there was no evidence of plant survival and where the volcanic substrate contained no residual organic material, the revegetation process was much slower. Aston (1916) also attributed the fast recovery of vegetation on the Tarawera Mountain eruption site (New Zealand) in part to hold-over plants. Griggs (1918, 1922) reported that at Mount Katmai, Alaska, the first plants to appear originated from buried root systems. Some of these, such as Equisetum arvense, were found to grow through 30-36 inches of ash. The hold-over plant recovery was so rapid in areas of ash fallout that he had to transfer his study to more deeply-buried sites. Griggs (1933), in particular, stressed the importance of distinguishing between hold-over plants and new colonizers.

2. Upon complete destruction of existing plant communities, seed sources for new colonization may become a limiting factor (Rigg 1914; Eggler 1963) and new substrates may lack organic nitrogen entirely (Griggs 1933; Tezuka 1961) or the level of organic nitrogen may be very low (Eggler 1963). From these studies it becomes apparent that the type of colonizers may be in part a function of the destruction effects.

Griggs (1933) found that the first colonizers on the deep ash deposits on Katmai were members of the Jungermanniaceae (liverworts), which apparently can grow on substrates almost devoid of organic nitrogen. Mosses and algae invaded only after nitrogen levels had increased. A local increase of nitrogen levels on new materials was observed beneath mosses (Eggler 1963; Tagawa 1964). In some tropical habitats the first colonizers have been algae (Treub 1888; Booth 1941; Dory 1961, 1967b). However, there are obviously great differences with regard to species, population density, and habitat. Some investigators thought that algae may supply nitrogen for higher plants. Treub observed thick carpets of blue-green algae on the new substrates of Krakatoa. He believed that these Cyanophyceae were the first plant colonizers forming the germination medium for the establishment of ferns. Booth (1941) reports that algae were among the first colonizers of eroded soils in the south-central United States. He found a complete covering of Myxophyceae that extended over hundreds of acres. When Brown et al. (1917) found no algae on the volcanic soils of Mt. Taal, they concluded that this condition explained the paucity of vascular species. In contrast, Dory (1967b) observed that the initial colonizing algae had little or no effect in aiding the establishment of higher plants.

Some investigators (e.g., Treub 1888) have contended that higher plants colonizing new volcanic substrates must get their nitrogen from sources other than nitrogen-fixing algae. Ernst (1908) and Campbell (1909), in continuing the Krakatoa study, found an aerobic nitrogen-fixing bacterium in the new soil. They also found abundant nitrogen-fixing bacteria in the root nodules of several leguminous plants on new volcanic substrates. Campbell assumed that bacteria were among the first pioneer colonizers, and that their presence helped to establish higher plants by providing organic nitrogen. Gates (1914) reported that a leguminous shrub, Acacia farnesiana, quickly became established on the new sterile soil of Taal Volcano, because the plant was associated with nitrogen-fixing bacteria.

Apparently, algae do not play such an important role in other tropical habitats. Hasselo and Swarbrick (1960), studying a section of the 1959 lava flow on Cameroons Mountain, found that creeping herbs that were rooted in undisturbed neighboring soil were the first colonizers. While they noted no definite algal stage, they found algae covered about 20% of the surface. They also drew attention to seasonal variations which were shown particularly by mosses and herbaceous plants. Tagawa (1964), working on Sakurajima, found early development characterized by bryophytes and lichens and he drew attention to the invasional differences between bryophytes, lichens, and higher plants; the cryptogams have more universal means of distribution and the higher plants are more directionally distributed from seed source centers. Tezuka (1961) found that mosses and lichens were not important on Oshima and that their role as pioneer invaders had been exaggerated by Clements. He also could not recognize the so-called "herbage" stage. However, a distinct herbaceous stage was recognized by Tagawa (1964) on Sakurajima. Also, Millener (1953) claimed that the classical sequence of algae-moss-fern-higher plant arrival was not applicable in many habitats on Rangitoto Volcano (New Zealand). Instead, he claimed that woody plants, Metrosideros excelsa, were pioneer colonizers. He explained the mode of pioneer colonization by woody plants as expansion of islands of forests that became established as circular colonizers in various places on an otherwise uniform lava surface. Millener (1965) also observed that of the 400 species of vascular plants found on the lava flows more than half were non-natives.

Similar to Forbes' (1912) and Skottsberg's (1941) findings on the island of Hawaii, a fern stage was recognized by Keay (1959) on Cameroons Mountain, which was well established 14 years after deposition of the 1922 lava flow in a very wet region. Shrubs were established 29 years after the eruption, while ferns persisted. Keay reported 12 species of Ficus as pioneer trees.

Tagawa (1966, 1968) also studied disseminule dispersal with cotton traps on Sakurajima. He found that several pioneer mosses were wind-dispersed by shoot fragments that would produce new plants. Spores were produced only during dry periods. The main reproductive mode appeared to be vegetative. He also noted that where topography influenced wind movement, this factor was important in disseminule dispersal, perhaps even more so than the direct distance to the seed source.

3. Different rates of invasion with regard to differences in substrates were observed by Eggler (1963) on El Paricutin. Plants were able to start on lava but not on unmodified ash. Mosses started in lava cracks where run-off water accumulated. Invasion rates were related to accumulation of wind-deposited ash that had sifted into cracks rather than to the differences of a'a and pahoehoe lava. This observation is similar to that of Skottsberg (1941) on Hawaii. Eggler also observed faster colonization of mosses where soil water was locally augmented by condensation near steam vents. Earlier, however, Eggler (1941) noted a slower vegetation succession on a'a lava in southern Idaho, as did Forbes (1912) on Hawaii. Tagawa (1964) also found more mosses and lichens on rough lava (a'a) than on smooth at Sakurajima. Taylor (1957) found ash-particle size to be one of the major factors causing differences in plant establishment on recent volcanic deposits in New Guinea (Mt. Lamington, Waiowa Volcano, Mt. Victory). Plants tended to get started earlier and increase more in cover on thin ash deposits than on deep ones. He attributed this to the finer material that was sorted out at the margin of the ash blanket. Here, the fine material was assumed to hold more water than the coarser ash in the deeper parts. Similarly, Treub (1888) described the deep Krakatoa pumice material as a very dry substrate. Dilmy (1965) observed that the hard-crusted soil on the 1963 Agung Volcano eruption site impeded new plant establishment. Survival plants were among the first to appear in the devastated area. However, both these and pioneer plants could only become established along small rivers and on moist sites.

Schwabe (1969) and Behre and Schwabe (1969) reported briefly about the initial plant colonization of Surtsey, the volcanic island near Iceland that surfaced in 1963. Activity had ceased in 1967. Within one year after that, they noted a dense cryptogamic flora in and around fumarolic areas. They listed eight species of blue-green algae (one of them a nitrogen-fixer), three species of mosses, and a large number of diatoms. This initial colonization was associated particularly with small ash mounds near fumaroles. The vapor steam of the fumaroles had apparently caused the formation of these miniature dunes by binding fine, air-blown ash particles through moisture condensation. The authors refer to these locally favorable substrates as "oases" and say that they make up about 13% of the total new surface area. Occasional germinants of herbaceous seed plants were observed near the coastal points where seed was apparently inadvertently introduced with supplies. However, none of these higher plants have exhibited a community-type establishment.

The analysis of previous studies has shown that although some significant invasion patterns have been recognized, there are still gaps in the knowledge. For example, algae always seem to be among the first invaders, particularly where there are no surviving plants. But their significance in terms of abundance and interaction with other life forms has been evaluated very differently. In some studies algae are believed to provide the necessary organic nitrogen for higher plants or to form at least an important germination medium for these, while from other studies the role of algae seems completely independent and without relation to the other plant-life forms. Lichens and bryophytes appear to be significant only in certain climates. Ferns seem even further restricted because their presence is not often emphasized outside the tropics. Vascular plants can be among the first pioneers, but their appearance seems to be related to the degree of volcanic destruction, which makes their role as "true" pioneers somewhat questionable.

The review shows also that it seems difficult to define habitats before any definite community pattern is established. This is reflected in the rather general descriptions of volcanic environments.

Long-term observations of the same areas are fragmentary at best. This relates to the listing of species and life forms as well as the assessment of associated habitat factors. Very little emphasis has been placed on the relationship of life forms to one another and to the habitat complex. Changes in vegetation structure are only vaguely documented. Photographic records of the same volcanic habitats over a period of time are almost unavailable. There seems to be a near total lack of systematic and quantitative records. While species lists have been published for certain pioneer and later successional stages, there seem to be no continuous records of increase and fluctuation in species numbers in the same area through a sequence of years.