Aloha Friends of Hakalau Forest 

On the facilities end of things, the volunteer cabin got a facelift to the exterior. For those who has stayed in the volunteer cabin over the years you have had a front row seat to watching its transformation into what we now see. Both the refuge and the FoHF have put a lot of hours and dollars into the facility and it is one that is close to our heart. Recently, the exterior decks were removed and replaced with a single, unified wrap around deck on the side by the garage. Additionally, staff installed a new shower in the bathroom.

 The clearing of forests for cattle ranching is one of the major causes of tropical forest loss worldwide.The high degree of isolation of the Hawaiian Islands has resulted in over 5 million years of plant evolution in the absence of mammalian herbivores and extreme levels of endemism in most plant taxa. The impacts of cattle and other ungulates on the tropical forests of these islands are well documented and exemplify how large, non-native mammalian herbivores can modify tropical island ecosystems through herbivory, girdling, rooting, and trampling. Because of these impacts, forest restoration projects throughout the state of Hawai’i usually begin with fence construction to exclude non-native ungulates. Unfortunately, most wet forests below approximately 1,500 m often contain populations of one to several highly invasive, non-native shrub and small tree species such as Clidemia and Strawberry guava. The disturbance caused by ungulates, particularly pigs, facilitates establishment and spread of these invasive plant species in native forests, which can greatly complicate restoration attempts because these species may also respond positively to the removal of ungulates. Thus, passive forest restoration, whereby the forest is allowed to recover unassisted following the initial restoration action is often simply not possible in Hawai’i. Instead, active restoration techniques such as regular removal of invasive plant species are required, at least initially. Because these techniques are labor intensive and expensive, they are often not successful in the long run.

 Cattle were introduced to the island of Hawai’i by G. Vancouver in 1793, and their numbers increased rapidly, particularly in the windward montane forests of Maunakea and Kohala volcanoes. Through the 20th century, ranchers maintained large areas of former forest for pasture and in other still-forested areas allowed cattle to graze without active management, a practice that continues to the present time. In 1985, Hakalau Forest National Wildlife Refuge was established to preserve approximately 13,000 ha of montane wet forest that had been impacted by cattle since the early 1800s and later by feral pigs. Some of the first management actions were to fence large tracts of the preserve and to actively remove all remaining cattle, most of which had become feral. All cattle were removed by 1992, and most feral pigs were removed from over 4,000 ha of fenced tracts by the early 2000s.

 The approximately 150 years of cattle grazing, and to a lesser degree feral pig herbivory and rooting, left much of the formerly diverse montane wet forest a woodland with large old remnant trees comprising the canopy and an understory of primarily introduced pasture grasses with scattered remnant native shrubs and small trees growing primarily as epiphytes or on nurse logs (Hart 2010). Nonnative trees such as strawberry guava were notably absent. A primary goal of refuge managers was to allow the forest to passively regenerate, mainly to improve habitat for a number of remaining threatened and endangered native Hawaiian forest bird species, but also to increase native plant diversity and native Hawaiian forest cover. To what extent have these passive restoration efforts been successful? In this study, we characterize the resiliency and examine patterns of regeneration in a montane Hawaiian wet forest in which invasive tree species are absent and that has been passively managed through fencing and ungulate removal. In particular, we characterize recruitment within long-term plots and measure changes in abundance and basal area (BA) (the area of a trunk at breast height) for more than 7,100 marked trees over a 17–18-year period. Also, because large, “oldgrowth” trees play a disproportionate role in the ecology of many rare Hawaiian forest birds we address how large-tree abundance and dynamics have changed in this forest

Hakalau Forest National Wildlife Refuge (Hakalau) is on the eastern slope of Maunakea volcano on the island of Hawai’i. The section of Hakalau where this study occurred consists of a moderately disturbed mix of open and closed forest dominated by ‘ōhi’a. These trees have a fairly low mean annual growth rate of 1.3 mm/year, commonly grow to 1 m or more in diameter, and may reach ages up to 600–700 years (Hart 2010). Koa is codominant with ‘ōhi’a in some areas and grows approximately three times faster, but is far less abundant. Numerous other small trees or large shrubs occur in the mid-canopy of this forest, including kōlea, ’ōlapa, kāwa’u, pukiawe, and pilo. The growth rates for these mid-canopy trees ranges from a mean of 1.1 mm/year for kōlea to 3.2 mm/year for ’ōlapa (Hart 2010). The mid-canopy also includes tree ferns but this species is relatively rare in the higher elevation forests included in this study, possibly due to the long history of ungulate disturbance. Introduced pasture grasses continue to dominate the ground cover in many areas. The open forest where our study took place is bounded at lower elevations (below approximately 1,500 m elevation) by denser, less disturbed wet forest, and at higher elevations by an area that was previously deforested and converted to grassland, but is now being actively reforested with koa trees.

 In 1996, two 100 ha study sites, Pua Akala (upa) and Pedro (ped), separated by approximately 4 km of forest were established at elevations ranging from 1,750 to 1900 m. Both study sites were within fenced areas from which all cattle and nearly all pigs had been removed by 1992. Within each study site, ten 1-km-long transects were established with permanent 30 m diameter circular plots (stations) placed at 100 m intervals (200 plots total). From 1996 to 1997, all stems within each plot ≥5 cm diameter at breast height (dbh) were marked with an aluminum tag secured to the tree with an aluminum nail at approximately 1.3 m above the ground. The diameter of each tree (both live and dead) was measured at a point approximately 3 cm above the tag, for a total of over 7,100 trees. From 2014 to 2015, all tagged trees within each of the survey plots were remeasured at the same location on the tree, and changes in growth state (live to live vs. live to dead) were determined. All newly recruited trees ≥5 cm dbh were also tagged and measured. Large trees were classified as ‘ōhi’a or koa individuals ≥50 cm dbh (following Hart 2010).

We first analyzed on a landscape scale (i.e. grouping all plots together) the rates of population changes between the two survey periods. Annual rates for net change, and recruitment were estimated. This was done for each species separately and for all species grouped together.

BA for each tree was calculated as the area of the trunk at breast height [π(D/2) ²]. BA across the landscape can increase through the growth of already established trees and through the recruitment of new individuals into the population. BA can decrease over time primarily through mortality. BA changes were calculated as the difference in total BA for each species between the two time periods. We examined how the proportional representation of individuals ≥5 cm dbh may have changed over time for each species. We also looked at differences in annual mortality rates for small (5-50cm dbh) and large size classes of koa and ‘ōhi’a.

There was an overall increase in abundance of each of the seven tree species (≥5 cm dbh) censused during the 1996–2015 survey period). ’Ōlapa trees increased by approximately 800% and kāwa’u and pilo by close to 500%. While ‘ōhi’a were the most abundant species overall, they only increased in abundance by 16.5%. Annual rates of recruitment and mortality varied both by species and site (Pua Akala vs. Pedro) with many of the rarest species from the original survey such as pilo, koa, and kāwa’u having the greatest annual increase in recruitment (Fig. 1). Recruitment and mortality for, kāwa’u, kōlea, ’ōlapa, and pukiawe were similar among sites, while ‘ōhi’a, which was by far the most abundant species overall, had much higher recruitment at Pedro than Pua Akala (Fig. 1).

There were also large changes in the size class distribution for most species between surveys (Fig. 2). Most species exhibited a deficit in the number of small trees ( ˃10 cm dbh) trees compared to larger ones during the first survey. This deficit was particularly important for koa as most of the individuals inventoried were larger than 50 cm. This deficit had largely been erased by the time of the second survey. Overall, the number of small trees was much greater during the second survey and tree size class distributions tended to follow reverse J-shaped curves (Fig. 2). The number of large koa considerably decreased while the number of large ‘ōhi’a slightly increased.

This study demonstrates that in the absence of ungulates and invasive woody plant species, tree populations in Hawaiian montane wet forests are relatively resilient following long periods of ungulate disturbance.

 The substantial changes in tree size distributions for each species were a notable outcome of this study. Tree size class distributions are commonly used as a measure of forest health and are often associated with population trends. Populations with few juveniles and size class distributions that are skewed toward larger individuals tend to be a sign of a forest in decline. Conversely, when there are large numbers of juveniles in the population and fewer large individuals, the size class distribution of trees resembles a reverse-J shape and is one of the factors indicative of a healthy old-growth forest. During first survey that was conducted soon after ungulates were excluded from the refuge, the size class distribution of most species was relatively flat or even skewed toward larger size classes (e.g. koa), indicating a lack of recruitment for most species in the decades leading up to this study. By the second survey, however, the size class distribution for most species better resembled a reverse-J shape with large pulses of new recruits entering the smaller size classes. The relatively fast reversal of the size class distribution curves demonstrated here provides further evidence that Hawaiian forests long impacted by ungulates can regenerate following their removal and in the absence of invasive tree species.

The high annual mortality of large koa trees at Pua Akala with little sign of recruitment into this size category since the mid-1990s provides evidence that the proportional representation of koa versus ‘ōhi’a in the forest canopy at this site and likely other areas within Hakalau has been changing for decades. Most of the large koa that persist at Hakalau were probably fairly small but too big to be eaten by cattle as they began to invade this region by the mid-1850s (and likely earlier). With size-specific growth rate estimates indicating that koa rarely reach ages greater than 200–250 years in this forest (P. Hart unpublished data), this population has likely been in a long period of senescence and decline. If we assume that the 2.7% annual mortality for large trees for the two sites combined demonstrated here has been similar for this koa population in the decades leading up to the initial survey, then the forest canopy in the past was likely a more even mix of koa and ‘ōhi’a trees than it is now. Hawaiian montane wet forests have been disturbed to such a point that understanding the past vegetation structure and community composition in the absence of disturbance remains challenging. However, the pulse of new koa recruits indicates that there will likely be a more even balance of koa trees in the larger size classes in the coming decades at Hakalau.

The apparent decline in large koa density at Hakalau has important implications for the native Hawaiian bird community. Large koa trees acquire an architecture that makes them important as nest sites for the obligate cavity nesting, federally endangered Hawai’i ’Akepa. Nest site limitation has been implicated as a factor responsible for “anomalous distributions” of ’Akepa across the landscape at Hakalau. Large koa trees are also the primary foraging habitat for the federally listed and highly endangered ’Akiapōlā’au. While this insectivorous honeycreeper species is known to forage on other native tree species when available, they forage almost exclusively on koa at Hakalau and their population size and territory size appears to be tied to the abundance of koa across the landscape. Given the apparent decline of koa at Hakalau over recent decades, it is likely that the populations of ’Akiapōlā’au, ’Akepa, and other species that utilize koa have declined as well. The exclusion of cattle and pigs from the refuge and the resulting pulse of new koa, as well as ’ōlapa, kōlea and other recruits demonstrated here indicate that the resources for these birds may be regenerating just in time to prevent their further decline on the refuge.

We’ve all heard of “seat of the pants” research. Sometimes things just have to happen that way. Soon after Hakalau Forest National Wildlife Refuge was established, one of the main projects was to reforest almost 5,000 acres of upper pasture lands at the highest elevations of the refuge. This area , between 6,000 and 6,500 feet. had been a rich Koa forest over 100 years ago, but because of cattle grazing it had been reduced to open grasslands. Dick Wass, the first Refuge Manager for Hakalau Forest NWR, and I had started working with volunteer groups to assist in the reforestation effort in 1989. These volunteers planted Koa seedlings near the refuge upper boundary as well as long lines of plantings down towards the old growth forest at lower elevation. The strategy at that time was to plant Koa bands, or lines of Koa, three to four trees wide about 12 feet apart to emulate the mauka-makai tree lined gulches that traversed the upper grassland pastures. It had been noted that native birds often followed the trees along these gulches, flying from tree to tree, uphill, and could be heard regularly even near the higher elevation Refuge Volunteer Cabin far from the old growth forest. Cattle were unable to graze in these deep, steep walled gulches, allowing native trees and shrubs to grow along the gulch walls. Birds such as Oma’o carried seeds from the forest below to these upper areas allowing the forested gulches to continue to provide habitat for forest birds. By planting Koa in mauka-makai bands, it was thought that as the Koa grew taller the birds would follow the planted Koa bands spreading native plant seeds and these trees and shrub species would eventually spread uphill over time.

To keep track of the Koa plantings, several times each year we would measure the young Koa to determine survival and growth rates. Early on we noticed that the survival rates at the higher elevations were a lot lower (42%) than those at lower elevations (up to 100%). When checking the upper elevation Koa saplings in the spring, we often saw that many of the Koa saplings had been burned or killed by frost during the winter months. This was definitely a problem that had to be dealt with if we were going to reforest the upper elevations of Hakalau Forest NWR.

To deal with the frost issue, we decided to look at possible frost preventative methods. No one in Hawaii had dealt with this issue before, and there was scant information in the literature. Not knowing exactly what to do, Dick and I brainstormed to see if we could come up with anything that we thought might work to deter the plants from freezing or getting frost burned. It was noted that farms on the mainland used various bark and sheet-plastic mulches, as well as “Wall O waters” or “Tubex” to keep frost at bay, so these ideas were a good place to start. We came up with ten “methods” and set out to test them. They were: 1. “Wall O waters”, a small commercially available 18 inch tall “teepee” with walls filled with water that heated by the sun during the day and protected the young Koa from freezing at night; 2. A 3 foot tall, two foot diameter cylinder of hogwire fence covered with clear sheet- plastic; 3. A similar sized cylinder covered with shade cloth; 4.  A similar sized cylinder with shade cloth covering only the downhill or east side; 5. A similar sized cylinder covered with shade cloth only on the uphill or west side; 6. A similar sized cylinder with only the top covered with shade cloth; 7. Tubex, a commercially available translucent plastic tube about 2 ft tall, 5 inches in diameter placed over the plant; 8. A 3 foot square of sheet-plastic “mulch” around the base of the plant; 9. A 6 inch thick layer, 2 foot in diameter of redwood bark mulch around the base of the plant; and 10. Lastly a 3 foot tall hogwire fence, two foot diameter cylinder with no covering at all as a control.

We set up ten of each of these “devices” around 100 randomly selected newly planted Koa in the fall of 1993 at one of the highest elevation planting bands about 6,500 ft elevation. This was one of the most frost prone areas and elevations. Hoping for the best, we were really not sure what to expect. This was definitely a “seat of the pants” test.

The following spring we surveyed our plantings. Right away we saw that some methods/devices didn’t work well at all. They had high mortality and frost burn (1,7, 8, 9, 10). Other devices did have partial success, but they were very time consuming to construct, transport, deploy and maintain (2, 3, 5, 6,). For example, can you imagine spending hours cutting and constructing a hog wire fence tube 3 ft tall by 2 feet diameter, then attaching clear plastic sheeting covering the whole apparatus. Hundreds of these would need to be carried out to the planting area and set over a plant. Two fence posts would also be needed and attached to the device to hold it upright and anchor it to the ground to keep it from blowing away in heavy wind. Multiply this by hundreds or thousands of tubes and you can understand the great cost and time expended per seedling if this method was used.

So which method was the winner? It was #4. a hog wire fence tube with shade cloth on the east side. This method provided 100% survival of the young Koa saplings. At the time, we had no idea why it worked, but there it was, no mortality. To reduce construction time and cost we ended up changing the design a bit to make it easier to construct and deploy. Instead of hogwire fence, we used a piece of shade cloth 3 feet tall by 2 feet wide stretched between two wooden stakes driven into the ground to hold the shade cloth like a wall on the east side of the Koa seedling. They were easy to construct, a bundle of ten could easily be carried by a volunteer and all that was need for deployment was a small sledge hammer to pound the two posts in the ground.

Next was “phase two”. We needed to determine if this method really and truly worked.  With the help of volunteers, we constructed 100 stake and shade cloth “barriers” and set them on the east side of 100 randomly selected Koa seedlings in an area where 1000 trees had been planted in the fall of 1994. This plot was only a few hundred yards above or west of the refuge cabin. In the spring of 1995 we surveyed these plantings. It was hard to believe our eyes but 100% of the test plantings had survived while only about 40% of the other plantings survived. Now we had our method that prevented frost mortality of young Koa trees! The new devices became known as FPD’s for “Frost Protection Devices”.  They were relatively cheap to make, costing only a few dollars each. Refuge volunteers could easily construct and deploy 500 to a 1,000 of them on a weekend. The perfect solution to a big problem.

How do they work? The US Forest Service and UH Manoa experimentally tested the young Koa and the FPDs and determined that the young Koa can handle the freezing process but the rapid thaw by the early morning sun was detrimental. Rapid defrost caused the cell walls to rupture leading to desiccation. The FPD shades the plant from the early morning sun allowing the ice crystals inside the seedling to melt more slowly preventing rupture of the cell walls. Also, at night when the temperatures are the coldest, the shade cloth reflects the heat from the ground back onto the plant, so the plants don’t get as cold as when it’s in the open. And, as a further benefit, the shade cloth condenses moisture from fog and deposits it next to the young Koa providing more moisture. All these benefits are from a small shade cloth wall on the east side of the young plant. Furthermore, we discovered that when the young plants grow to about 3 feet tall, a year later, they are no longer prone to frost damage. At this point the FPD’s can be removed and reused on the next year’s planting.

So, the next time you are at Hakalau Forest NWR and driving down the road to the Volunteer Cabin, watch for a brown sign on the left, hidden in the trees, commemorating the planting of Koa trees sponsored by the Global Relief Fund. This is the “Phase Two” test planting area. Stop, listen to the birds, look at the tall thick Koa forest in front of you, thanks to the innovative FPD. Only twenty- five years ago this was an open treeless pasture with no native birds. Change is what Hakalau Forest NWR is all about.

Every Wednesday on HPR listen to Manu Minute created by Patrick Hart, member of the Friends since its inception and a board for member for multiple terms. Click the button to hear the segments that have already been aired.

James Quinn and Florence Hoffman should have been in the $1000 to $4999 donation category

Donald Kanakawaiwai Romero was misspelled.