An Ecological Model of the Effects of Exotic Factors on Limiting Hawaiian Honeycreeper Populations1 STEPHEN MOUNTAINSPRING, Mauna Loa Field Station, P.O. Box 88, Hawaii National Park, HI 96718 ABSTRACT. The object of this study was to construct a quantitative model to compare the effect of several exotic factors on limiting natural populations. The effect of habitat loss, range loss due to avian disease, feral ungulate activity, and interspecific competition for food resources from exotic birds on 33 populations of 16 Hawaiian honeycreeper species was examined. The impact of the four limiting factors was characterized in terms of the percent reduction in range or population density. Habitat loss resulted in a 74% average reduction in the original range; probable range loss due to disease averaged 46% of the available habitat; and feral ungulates and exotic birds lowered honeycreeper population densities by an average of 22% and 9%, re- spectively. The combined effect of the four limiting factors, which was modelled as their multiplicative product, represented an average loss of 89% in potential population size due to their operation, and accounted for 75% of the variation in present population sizes. OHIO J. SCI. 86 (3): 095-100, 1986 INTRODUCTION ral populations of habitat loss, avian disease, feral pig (Sus The relative importance of factors that limit the size of scrofa) activity, and food competition from the Japanese animal populations is a special concern in wildlife man- White-eye (Zosterops japonicus) will be modelled quan- agement (Leopold 1933). Attempts to quantify the rela- titatively for the honeycreepers inhabiting the islands of tive roles of various factors have generally focused on Hawaii, Maui, Molokai, Lanai, Oahu, and Kauai. detailed life history studies of individual species through Inclusion of the last two factors in the model is well techniques such as key factor analysis (Williamson 1971). justified. In the wetter forests that most honeycreepers In this paper I will develop a quantitative model inhabit, the chief habitat modifier at present is the feral to compare the relative effects of limiting factors of ex- pig, whose rooting and wallowing activities drastically otic origin on populations of the Hawaiian honey- modify understory and eventually canopy composition creepers (Drepanidinae). (Tisdell 1982). The analysis of interspecific competition The Hawaiian honeycreepers have a spectacular range for food resources focuses on the Japanese White-eye of morphological and behavioral adaptations to feed on because this abundant and omnivorous species appears to foliage insects, timber-boring insects, nectar, fruits, have the greatest impact of exotic birds on honeycreeper seeds, molluscs, and seabird eggs (Amadon 1950, Berger populations (Mountainspring and Scott 1985). 1981). Unfortunately, excessive perturbation from exotic THE QUANTITATIVE MODEL elements to native Hawaiian ecosystems has created the highest concentration of endangered birds in the world Shelford (1913) was the first to articulate the Law (King 1978). Extant honeycreepers (Table 1) represent of Tolerance, stating that given a series of factors that only 30% of the species present at Polynesian contact operate independently to reduce a population from its ca. 400 A.D. (Olson and James 1982); 14 of the potential, the combined effect of these factors can be 20 extant species are protected under the Endangered characterized as their multiplicative product. A similar Species Act of the United States. Due to limited fund- multidimensional conceptualization is found in the niche ing, identifying the relative roles that various exotic theory of Hutchinson (1958) and Whittaker et al. (1973). factors play in limiting honeycreeper populations is a This concept provides the basis for the mathematical crucial step in efficiently managing the recovery of en- development below. dangered species. Let Fx be the effect of factor 1 on reducing the popu- Numerous exotic factors have been suggested as con- lation of a species. A value of 0.1 would signify that the tributing to the decline of honeycreeper populations, population is reduced 10% due to the operation of including habitat loss due to human activity (Olson and factor 1. Let F2..N represent the effect of factors 2 through James 1982, Berger 1981), gradual habitat degradation N, with each Fk constructed so as to account for the due to feral ungulate activity (Scott et al. 1986), avian operation of Fu . ,Fk-i (i.e., the factors are ortho- disease (Warner 1968, van Riper et al. 1982), food com- gonal). The fraction of the population remaining after petition from exotic birds (Mountainspring and Scott accounting for the operation of Fk is thus 1 — Fk. The 1985), predation from rodents and carnivores (Perkins fraction of the potential population remaining after the 1903, Atkinson 1977), invertebrate predators on the food combined effect of all factors is considered will be termed base (Banko and Banko 1976), exotic plant invaders the Remaining Potential Population (RPP). This value is (Smith 1985, Scott et al. 1986), and wildfire in dry computed as the percent of the original population pre- habitats (Tomich 1971). In this paper the effect on natu- dicted to remain to permit comparison between species with different carrying capacities and is given by: ^anuscript received 11 October 1985 and in revised form 24 Febru- ary 1986 (#85-48). 96 S. MOUNTAINSPRING Vol. 86 TABLE 1 Systematic arrangement (following Berger 1981) of the extant Hawaiian honeycreepers. Present habitat Forest type Elevation (m) Species Chief foods Laysan Finch*# Dry scrub 0-15 Insects, seeds, bird eggs (Telespyza cantans) Nihoa Finch*# Dry scrub 0-270 Insects, seeds, bird eggs {Telespyza ultima) Ou* Wet 1000-1500 Fruit (Psittirostra psittacea) Palila* Dry 2000-3000 Seed pods {Loxioides bailleui) Maui Parrotbill* Mesic to wet 1500-2000 Timber-boring insects {Pseudonestor xanthophrys) Common Amakihi Dry to wet 0-3000 Foliage insects, nectar (Hemignathus virens) Anianiau Mesic to wet 500-1500 Foliage insects {Hemignathus parvus) Kauai Akialoa*& Mesic to wet 1000-1500 Nectar, trunk insects {Hemignathus procerus) Nukupuu* Mesic to wet 1000-2000 Trunk insects, nectar {Hemignathus lucidus) Akiapolaau* Dry to mesic 1000-2500 Trunk insects {Hemignathus munroi) Kauai Creeper Mesic to wet 1000-1500 Trunk insects {Oreomystis bairdi) Hawaii Creeper* Mesic to wet 1000-2000 Trunk insects {Oreomystis mana) Maui Creeper Mesic to wet 1000-2200 Foliage insects {Paroreomyza montana) Molokai Creeper*& Wet 1000-1500 Foliage and trunk insects {Paroreomyza flammea) Oahu Creeper* Mesic to wet 1000-1300 Foliage insects {Paroreomyza maculata) Akepa* Mesic to wet 1000-2000 Foliage insects {Loxops coccineus) Iiwi Dry to wet 500-2500 Nectar, foliage insects {Vestiaria coccinea) Crested Honeycreeper* Wet 1500-2000 Nectar, foliage insects {Palmeria dolei) Apapane Dry to wet 0-2500 Nectar, foliage insects {Himatione sanguinea) Poo-uli* Wet 1500-2000 Molluscs, understory insects {Melamprosops phaeosoma) * Endangered species. # Found only on Northwestern Hawaiian Islands. & No sightings in past 20 years; may be extinct. The variance (J-2) of this measure is given by: sumed that lost or unoccupied suitable habitat would support bird densities equivalent to those in occupied areas, if the limiting factors were not in operation. Thus, for both binary and continuous variables, this scaling although in this study variances could be computed for represents the percent loss in potential population due to only some of the factors. the operation of that limiting factor. Because population size has two components, area and To validate the model (i.e., to make an assessment of density, the factors to be analyzed fall under two rubrics: its biological reasonableness) I examined the relationship those that are characterized as binary (all/none) in opera- between remaining potential populations as predicted by tion, and those that are continuous. The effects of habitat the model and actual present populations. This test as- loss and disease are modelled as generating sharp geo- sumes that present population sizes, which range from 30 graphic range boundaries; these are quantified in terms of to 1,100,000, will generally reflect the combined sever- the percent loss of range due to the operation of that ity of the four exotic factors. Using the product-moment factor. The effects of feral pigs and Japanese White-eyes correlation test (Steel and Torrie 1980) to examine various are modelled as modifying the population density of af- transformations (linear, logarithmic, square root, arc fected species at the scale that the data were collected (see sine), I found the best linear fit resulted from a loga- Wiens 1981); these are analyzed statistically to yield the rithmic transformation applied to population size and a percent difference in bird density due to their activities. square-root transformation applied to relative remaining To facilitate comparisons among the factors, it was as- potential population. However, the choice of trans- OhioJ. Science HAWAIIAN HONEYCREEPER LIMITING FACTORS 97 formation made little difference in the overall statistical the percent difference provided a statistical quantification accounting for variation in density of the bird and in values of the limiting factor. patterns observed. From standard regression analysis (Draper and Smith 1981), the predicted density at no effect (Bo) was calculated by: METHODS AND MATERIALS B0 = B-rXEX s(B)/s(E), Data