Search

Alien species are a major cause of species extinctions globally and the effects of introductions can become substantial, especially once aliens become entrenched and impossible to eradicate. Some entrenched aliens may assume niches of extinct native species that are functionally similar, but the capacity of an alien to perform specialised ecological processes is unknown. Here we examine the potential for the alien black rat (Rattus rattus) to effectively pollinate native Banksia ericifolia in heath at North Head, Sydney where endemic mammalian pollinators such as brown antechinus (Antechinus stuartii) and eastern pygmy-possums (Cercartetus nanus) had been locally extinct prior to recent reintroductions. Using artificial inflorescences, we compared the use of inflorescences, visit frequency and duration of alien black rats to reintroduced natives according to the stem diameter of the inflorescence. Black rats were the most frequent visitor and they spent similar or more time at artificial inflorescences as natives, but they also spent more time visiting inflorescences with thicker stem diameters. In contrast, the native mammals visited artificial inflorescences regardless of the stem thickness. We suggest that, as black rats are heavier than the native mammals, this could limit their capacity as substitute pollinators of B. ericifolia. This finding shows the importance of understanding the extent of the benefits of aliens before relying on them as functional replacements for extinct natives.

Factors that influence reproduction in nectar-feeding non-flying mammals are poorly described. We investigated factors that may influence the breeding traits of the eastern pygmy-possum (Cercartetus nanus), a small (25 g) nectar-feeding marsupial from eastern Australia. Females at our coastal site produced 1–3 litters (frequently of 5–6 young) over an 8-month period within the flowering period of the dominant food plant (Banksia ericifolia). The number of lactating females over time was highly correlated (R = 0.9) with the abundance of flowers on B. ericifolia, suggesting that flower availability has a substantial influence on breeding. To assess the generality of these findings and investigate whether elevation or latitude influence breeding in this species, we examined the breeding traits previously described at three other locations. Females on the tableland 70 km away produced one litter of 3–4 young over a 4-month period within the flowering period of B. ericifolia, the dominant food plant. At other coastal and tableland locations 500 km away, coastal females produced larger numbered litters more often than tableland females. A hypothesis relating to minimum temperature appears the most plausible explanation for this pattern that reflects elevation but not latitude. The mean minimum temperature drops below 5 °C for 2–5 months each year at the tableland locations but not at all at the coastal locations. Low temperatures are known to reduce nectar secretion in the dominant food plants. Thus, although the breeding traits of the eastern pygmy-possum are influenced by flowering in their dominant food plants, low temperature appears to impose a constraint on reproductive output.

Banksia ericifolia is killed by fire and so depends upon seeds for persistence. In contrast, B. oblongifolia can persist after fire by resprouting from protected buds, as well as by recruiting seedlings. These serotinous species coexist in the fire-prone scrublands of the Sydney Basin. Here we report experiments that compared seed release from cones retained in the canopy, seed germination, and seedling establishment in these two species. B. oblongifolia seeds were released more quickly than B. ericifolia seeds after a fire, and B. oblongifolia follicles opened and shed seeds after exposure to lower temperatures than those of B. ericifolia. B. oblongifolia seeds germinated more quickly than those of B. ericifolia. The relative growth rates and leaf mass ratios (mean ratio of leaf dry mass to plant dry mass over sequential harvests) of the two species did not differ. Root : shoot ratios were significantly large and biomass allocation to roots was faster in B. oblongifolia seedlings. B. ericifolia seedlings survived low and intermediate levels of droughting better than B. oblongifolia under greenhouse conditions. In a field experiment in which water was not limiting, large B. oblongifolia seedlings had the highest survival rate. These results suggest that selection has acted on B. ericifolia to minimize the chance of zero seedling survival by distributing the risks over more than one germination episode. There was no consistent evidence to support the hypothesis that selection on this species has increased its probability of seedling survival during any one germination opportunity relative to its congener.

Tap and primary lateral roots of seedlings of the putatively non-mycorrhizal Banksia ericifolia became marginally colonised when grown in an established mycelium of an arbuscular mycorrhizal (AM) fungus in the laboratory. A similar degree of colonisation was found in seedlings from an open woodland. All colonies lacked arbuscules. Two factors influencing colonisation and associated growth of host plants were examined experimentally: concentration of P in the soil and organic energy associated with the fungus. While some inoculated seedlings were slightly smaller when colonised by AM fungi, the results were inconsistent and never statistically significant. Seedlings take up insignificant quantities of soil P during early growth, even in the presence of abundant added P. Though colonisation was minor in all cases, an existing mycelium, whether or not connected to a companion plant, slightly increased the amount of root of B. ericifolia colonised by an AM fungus. All seedlings grew slowly. Shoots were significantly larger than roots, until the initiation of proteoid roots which commenced at about 40 days after germination, with both relatively high and low P supply.

Several features of the phenology and reproductive biology of members of the genus Banksia indicate that these plants either have a high level of mate choice or are obligate outcrossers, with many flowers being functionally male or sterile. These species produce many inflorescences with vast numbers of flowers, yet seed-set is typically very low. Most species studied appear to have at least limited self-compatibility and outcrossing rates close to one. Despite this, pollinators normally deposit a large amount of self-pollen on stigmas. In this study, we used experimental pollinations to test for evidence of preferential outcrossing within populations of Banksia spinulosa and Banksia ericifolia. Our allozyme analysis of progeny arrays indicates that the studied populations of both B. spinulosa from a prior study and B. ericifolia as determined for 27 plants in this study are highly outcrossed. Results from pollinations using applications of self- and cross-pollen and mixtures of the two indicated that, at ...

Several taxa of the Australian Proteaceae have invaded South African fynbos and require costly management programmes to bring under control. Banksia spp. have been introduced only recently to fynbos regions. The invasive potential of Banksia ericifolia (Proteaceae) was investigated by comparing its recruitment potential with that of an indigenous proteaceous shrub, Leucadendron laureolum. Both species are overstorey shrubs that are killed by fire and rely on canopy-stored seeds (serotiny) for recruitment. Eight year old B. ericifolia shrubs produced an average of 16 500 seeds per plant, which is thirty times more than the average of 570 seeds produced by 10 year old L. laureolum shrubs. The seed bank of B. ericifolia was not only larger than that of L. laureolum (1098 vs 525 viable seeds m−2 projected canopy cover), but also considerably larger than that described for the species in its native environment (200–330 seeds m−2 in a 9 year old stand north of Sydney). Leucadendron laureoleum released most of its seed a few days after the cones were burnt, whereas seed release in B. ericifolia was spread over 12 weeks. The seeds of B. ericifolia had lower wingloading and fall rates than L. laureolum and were dispersed over greater distances. The relative seedling growth rates of the two species were very similar (0.03 g per day), but below-ground biomass was greater and proteoid roots were more developed in B. ericifolia seedlings than in L. laureolum after 100 days. Four year old B. ericifolia plants growing in the field had attained over twice the height of indigenous pro-teoids and accumulated up to 10 times the fresh biomass of L. xanthoconus, a species which is ecologically similar to L. laureolum. The Bioclimatic Prediction System (BIOCLIM) was used to create a bioclimatic profile of B. ericifolia and identify climatically suitable areas in the Cape Province. Results show that its potential distribution covers most fynbos areas in the southwestern Cape. It is concluded that B. ericifolia has the potential to be highly invasive in fynbos.

I report observations of honeyeaters at sugarwater feeders conducted over two years in heathland near Sydney. Eight to 14 sugar-water feeders were distributed over a 4 ha patch of heath, and placed away from existing territories to minimise the chance of feeders being monopolised by a few birds. All common species of honeyeaters used feeders. and feeders were used extensively during all months except May-July (when nectar from Banksia ericifolia was abundant) and November-January (when there were few honeyeaters on the heath). Incidence of aggression at feeders was low, and most birds that were excluded could feed as soon as the aggressor finished using the feeder. Feeders were occasionally defended by large honeyeaters (Little Wattlebirds, Noisy Friarbirds and Red Wattlebirds) but these species were relatively uncommon. Feeders were never defended by New Holland or White-cheeked Honeyeaters, which were the most common species and defended breeding territories on the heath. Consequently, the majority of feeders were always freely accessible, so that they provided easy access to energy similar to that provided by rich and abundant nectar sources such as Banksia ericifolia. I conclude that feeders can be used to test whether correlations between nectar abundance and honeyeater abundance, behaviour and breeding are caused by responses to changes in energy availability.

This study examined the water relations of sclerophyllous evergreen vegetation (wallum) on coastal sand barriers in eastern Australia. Many wallum species may be groundwater dependent, although the extent of this dependency is largely unknown. Twenty-six perennial tree, shrub and herb species were investigated in three groundwater habitats (ridge, open depression, closed depression). Pre-dawn and midday shoot xylem water potentials (ψx) were measured monthly between late autumn 2010 and late summer 2011. Pressure–volume curve traits were determined in mid- to late spring 2009, including the osmotic potential at full (π100) and zero (π0) turgor, and bulk modulus of elasticity (ε). Carbon isotope ratios (δ13C) were also determined in mid- to late spring 2009, to measure water-use efficiency (WUE). The species displayed a range of physiological strategies in response to water relations, and these strategies overlapped among contrasting growth forms and habitats. Linear relationships between osmotic and elastic adjustment were significant. A strong correlation between δ13C and distribution along the hydrological gradient was not apparent. Banksia ericifolia subsp. macrantha (A.S.George) A.S.George, Eucalyptus racemosa Cav. subsp. racemosa and Eucalyptus robusta Sm. displayed little seasonal variation in ψx and maintained a comparatively high pre-dawn ψx, and are therefore likely to be phreatophytic. Wetland vegetation in the lowest part of the landscape appeared to tolerate extreme fluctuations in water availability linked to a prevailing climatic pattern of variable and unreliable seasonal rainfall.

We tested the ability to predict plant (and hence population) age for three fire-sensitive obligate-seeder Proteaceae species (Banksia ericifolia, Banksia marginata and Petrophile pulchella) in the heath and woodland vegetation of the Sydney region. To do this we sampled the number of growth whorls, as well as other growth characteristics (stem girth and height measurements, and canopy area and volume estimates), in areas of known time since last fire (TSLF). The average number of growth whorls was a very good predictor of plant age for both Banksia species (R2 = 98%, 99%), but this needed to be corrected for linear underestimation in P. pulchella (R2 = 92%). This technique could successfully be applied to these species in similar habitats across a large spatial scale, and so this information can be used to determine the age of a population in areas of unknown TSLF. A sample size of 15 plants was sufficient for accurate age estimates of all species; however, better estimates of TSLF for a particular plant community were obtained when estimates from two or more of the species were combined. We thus provide empirical evidence for the validity and accuracy of the growth-whorl technique for predicting plant age and hence TSLF. This information will assist in informing the development of appropriate management strategies for plants in relation to fire. Of the other growth characteristics studied, stem girth was the most reliable predictor; however, in general these other characteristics had wide confidence intervals on the predictions for sites greater than 10 years TSLF, owing to a non-linear relationship with age.

Banksia species are conspicuous in Australian heathlands and woodlands and provide a major source of nectar for a range of vertebrate and invertebrate pollinators. However, nectar production by Banksia is difficult to measure because inflorescences comprise large numbers (hundreds to thousands) of tiny flowers. Our aim was to quantify variation in nectar volume among B. ericifolia L.f. plants, but existing techniques were too time consuming or required destructive sampling of inflorescences. We estimated nectar volume by counting the number and estimating the size of nectar drops on a sample of flowers on each inflorescence. Our technique was fast (10 min per inflorescence) and a highly significant positive relationship was found between visual estimations and direct measures (capillary tubes and micropipettes) of nectar volume (r = 0.92; P ≤ 0.0001). This technique thus provides an alternative to current techniques for measuring nectar volume in plants that have large numbers of small flowers arranged in dense inflorescences. By using this technique, we found significant variation among days in nectar volume produced by each of five B. ericifolia plants (P ≤ 0.02). Furthermore, days of high and low nectar production were synchronised among plants. We concluded that the temporal variation found may be the result of fluctuations in environmental variables, rather than the age of an inflorescence. Indeed, nectar volume was found to be significantly correlated with a range of weather variables including daily maximum temperature (°C), precipitation (mm), wind speed (km h–1) and relative humidity (%). However, larger-scale and longer-term studies are needed to adequately examine any relationship between environmental variables and nectar production.

The effect of high frequency fire regimes on a coastal heathland north of Sydney was examined. Plant community composition and species frequency and density (of a subset of species) were compared among sites burnt by either one, two or three fires in the period 1988-1990 inclusive. Constrained ordination indicated that number of fires had a significant effect on floristic composition. Site variables also had a significant effect on floristics and frequency and accounted for a larger proportion of variation than did number of fires. Analyses of deviance indicated that frequency in 13 species was unaffected by fire regimes or site factors. Site factors significantly affected frequency in 42 species. Frequency in seven species was significantly affected by fire regimes (no site effect). In six of these, frequency was lower in areas subjected to either two or three fires compared to one fire. Three of these species were obligate seeders. Frequency in the other species (a herbaceous resprouter) was significantly higher in the area subjected to two fires. There was no significant effect of fire on species richness. Densities of seven out of a subset of eight species were also significantly affected by fire (two obligate seeder spp.) and site factors (four spp.), Significantly lower densities were related to multiple fires. The study indicated that regimes of frequent fire can deplete populations of some heathland species with the dominant shrub species, Banksia ericifolia L.f., being most affected.

Simulations were used to investigate the effect of season of fire on seedling emergence in Banksia ericifolia and B. serrata in the Sydney region, New South Wales. The simulations were based on models of soil-surface moisture in response to rainfall, seedling emergence response to soil moisture and post-fire seed release from fruits as determined by fire intensity, derived from field and laboratory studies. Fires were modelled on the first day of each calendar month for a 50-year period. Levels of post-fire seedling emergence were calculated using rainfall data from the Sydney Observatory for the corresponding period (1931-1980). Trends in seedling emergence as a function of month of fire were examined. Alternative sets of simulations were performed to assess the effect of variations in post-dispersal seed mortality, fire intensity and induced summer dormancy. In both species, mean emergence was affected by season of fire only when a 10% per month level of post-dispersal seed mortality was simulated (there was no fire-season effect at lower mortality levels). Highest predicted emergence occurred after summer fires and lowest emergence after winter fires. A reduction in rate of seed release (lower intensity fire) and induced seed dormancy in summer also had a minor effect with respect to fire-season in B. ericifolia. Reported levels of post-dispersal seed mortality in Banksia species are often high, and therefore, the simulations suggested that there will be an effect of fire season on seedling emergence. However, given the high level of year to year variation in seasonal rainfall in the Sydney region, fire-season effects are not predictable in the short term. In the longer term, the timing of fire relative to sequences of wet and dry years may be of equal importance to season of fire in its effect on populations of these species.

In the Sydney region, many plants from populations on windswept headlands have a more prostrate growth habit compared with plants from populations of the same species occurring away from the coast. To determine whether these different growth habits are genetically determined, plants from four populations of each of five species (Acacia rnyrtifolia, Acacia suaveolens, Banksia ericifolia, Casuarina distyla, Hakea teretifolia) were grown under uniform glasshouse conditions. Multivariate analyses of six morphological characteristics indicate that, for four of these species, the offspring are similar to their maternal parent; we thus conclude that the habit differences are genetically fixed in these populations. The same trend is apparent for

Observations and experiments were performed to examine the relationship between fire and seedling establishment in the following serotinous/woody species: Banksia ericifolia L.f., Banksia serrata L.f., Petrophile pulchella (Schrad.) R. Br. and Isopogon anemonifolius (Salisb.) Knight. For all species a moderate-intensity fire had no effect on the viability of seeds stored in cones compared to seeds stored in unburnt cones. Germinated seeds of all species, placed in meshed and unmeshed enclosures within an unburnt area, were consumed extensively by an unknown predator within 2 months of placement. No such effect occurred in similar treatments within a recently burnt area. Heating of the soil surface and the nature of substrates (litter, ash, bare soil) had variable effects on seedling establishment. Establishment of B. ericifolia and B. serrata was unaffected by the treatments whereas in P. pulchella and I. anemonifolius substrates and heating had significant effects. In P. pulchella establishment was significantly reduced on litter compared with pure ash. In I. Anemonifolius establishment was significantly higher on treatments with litter or mixed ash, except on soils that had received the low heat treatment, where there was no significant substrate effect. It was concluded that fire could create a more favourable soil-surface environment for establishment in P. pulchella only. In the other species, fire would have no effect or, in I. Anemonifolius, establishment might be enhanced in unburnt conditions. This conclusion does not take into account the effects of cover (overstorey and understorey) on establishment. It was concluded that predation of newly germinated seedlings may be the most important restriction on seedling establishment in unburnt populations.

Experiments were carefully designed to determine the breeding system of Banksia ericifolia L.f. An equivalent percentage of flowers (78%) contained pollen tubes following self-pollination and open-pollination while a significantly smaller percentage of flowers in an autogamy treatment (44%) and cross-pollination treatment (55%) contained pollen tubes. Significantly more of the inflorescences in the open-pollination (60%) and cross-pollination (33%) treatments produced fruit compared with those in the self-pollination (11%) and autogamy treatments (13%). We suggest that B. ericifolia is largely self-incompatible because fruit production did not reflect pollen tube abundances. The influence of pollination levels on fruit production was determined by reducing the number of flowers on an inflorescence to 100 (i.e. 10% of original) or fewer and hand-pollinating these with cross pollen. There was no difference in fruit production between inflorescences with reduced flower number and open-pollinated inflorescences which had their full complement of flowers. Thus, the ability of an inflorescence to produce fruit appears more likely to be determined by the type of pollen received (cross versus self) rather than by the number of pollinated flowers it contains.

Interactive effects of iron and phosphorus on the growth and appearance of Banksia ericifolia L.f. var. ericifolia seedlings were investigated. The seedlings were grown in a pinebark-based soil-less potting medium amended with 0-40 mg L-1 P added as single superphosphate and with Fe supplied either as FeSO4.7H2O (at 0-3 g L-1) or FeEDDHA (at 0-430 mg L-1). Fe deficiency was the dominant visual effect on shoots; its severity increased as P supply increased. Classic symptoms of P toxicity (death of oldest leaves) were produced in plants at the highest levels of P supply and low Fe supply. At each level of added P, Fe tended to lower %P in the shoots, mainly by allowing increased production of dry matter as Fe deficiency was overcome. The highest rate of FeSO4.7H2O addition reduced accumulation of P in the leaves and increased it in the stems. High P additions reduced accumulation of Fe into leaves. At non-toxic additions of P, young leaves had higher concentrations of P than did old leaves, but the reverse applied as P supply became excessive. The leaves of high quality plants contained less than about 0-14% P and more than 30 mg kg-1 Fe. P:Fe ratios for the leaves of such plants were in the range 14-53 (mean 23). Numbers of proteoid root clusters were greatest on plants in media of high Fe supply and low P and declined rapidly as P supply increased above the adequate range. No evidence was found for Fe stress leading to greater development of proteoid roots. Quality scores of at least 85% of maximum were mainly confined to media giving < 3 mg L-1 P in 2 mM DTPA (1:1.5 volume) extracts and amended with at least 1.5 g L-1 FeSO4.7H2O or 215 mg L-1 FeEDDHA. Response in ferrous sulfate treatments was better correlated with extractable P and Fe than with totals added, but there was no difference with FeEDDHA treatments, because additions of FeEDDHA did not decrease extractable P and ferrous sulfate did.

Plants of Banksia ericifolia and Petrophile pulchella are sensitive to fire. Changes in population size under different fire regimes were estimated, based on measurements of post-fire seedling emergence, seedling survival, survival and seed production in established plants of differing ages, survival of seeds held in serotinous cones and seed-release in the periods between fire. Seeds were first available at 5 years in P. pulchella and 6 years in B. ericifolia. Exact replacement would be possible when burnt at these ages, if seedling establishment were very high. Low establishment would delay replacement to 13 years of age in both species. Late summer/autumn fires of high intensity favour high establishment. Such fires at 8–10 year intervals would be tolerated without any sustained decline in numbers. Fires at 10–15 year intervals could occur regardless of season or intensity with little risk of a population decline. Large increases in numbers and density would follow fires spaced at 15–30 years. Enough seeds would be available for replacement up to about 50 years in both species. Viable seed-release in unburnt conditions was sufficient to compensate for deaths in stands over 20 years old, even with very low levels of establishment. Two variables accounted for the biggest changes in numbers and density between generations interspersed by fires; namely the age at which a stand is burned and the proportion of seeds which emerge as seedlings. A comparison with other similar species showed similarities in controls on emergence and establishment, lengths of primary juvenile periods and life spans.

A non-systemic Insecticide was used to exclude flower and seed predators from Banksia ericifolia and B. oblongifolia Inflorescenses. In B. ericifolia this treatment doubled the number of inflorescences that set seeds, and increased by 40% the number of seeds produced per Inflorescence. Insecticide treatment did not alter either of these components of seed-set in B. oblongifolia partly because the experiment began too late to exclude flower predators In this species. The results obtained for B. ericifolia support the hypothesis that flower and seed predators explain, in part, the low incidence of seed-bearing cones and the low seed:flower ratios reported for many Banksia species. Each Banksia species supports two guilds of Inflorescence predators. Two Lepidopteran species feed on young flowers; one of these causes tunnel damage to the rachis. Each Banksia species also supports five Lepidopteran and two Curcullonid seed-predators. About 40% of these herbivorous Insects are common to both shrubs.

High-density (dense) and low-density (sparse) plots were set up in naturally sown monospecific stands of Banksia ericifolia in coastal heath, 3 years after fire. This was done both in high-growth and low-growth areas. Plant mortality was recorded quarterly, and two harvests were made at 6 and 9 years to sample growth. Density-independent mortality at an exponential rate was observed in the low-growth treatments at both densities, and in the high-growth sparse treatment. Growth level affected mortality, with the half-life of populations in the high-growth sparse plots being double that of populations in the low-growth plots. Density-dependent mortality (self-thinning) was seen only in the high-growth dense plots. Seasonal effects on mortality were slight; maximum mortality was observed in the spring-summer period in plots subject to density-independent mortality, and in the winter-spring quarter in plots that had self-thinned. Yields in the high-growth plots and the low-growth dense plots were high for heath vegetation. The self-thinning populations did not exceed White's (1985) upper boundary for thinning lines of log intercept (K) = 5 on standardized axes. The data suggested a log intercept value in the range 4.8–4.9 in the high-growth stands assuming a thinning-line slope of – 1.5. Banksia ericifolia (a large shrub/small tree) has a high mean plant weight per given thinning density compared with trees, where an upper limit of log K= .4 has been suggested by White (1985). The volume of canopy space per plant in B. ericifolia is not unusual compared with other species. The amount of biomass packed into a given volume of canopy space was high in this Banksia, achieved by having leaves with a low ratio of area to weight (specific leaf area, SLA). For given values of density, leaf area index and proportion of shoot as leaf, plants with a low SLA will be several times heavier than plants with a high SLA. This achieves a high biomass to volume ratio without an erectophile canopy and may explain the high intercept seen for thinning lines of conifers.

Self-thinning B. ericifolia stands were observed to have high mean plant weights for their density, exceeding the upper boundary of the (log mean plant weight = 4.4–1.5 log mean plant density) line of White (1980) by approximately 0.5 units on the y scale (log mean weight per plant (g)). While proportion of shoot as stem and leaf area indices were similar in thinning stands of B. ericifolia and other species, specific leaf area was six to eight times less in the Banksia populations.

Fire intensity and seasonality affected the release of seed in Banksia ericifolia L.f, and the post-fire emergence of seedlings. Mean maximum temperatures recorded in a heathland burn in May 1977 at heights of 0.5, 2 and 3 m were about 400, 275 and 175°C respectively. Variability in maximum temperature was dependent on the age and position of B. ericifolia stands. Release of seed held on B. ericifolia bushes increased significantly after fire. The proportion of seed released on unburnt controls stayed constant. Seeds were released earlier and quicker from cones exposed to high fire temperature maxima than from those exposed to low fire temperature maxima. Up to 80% of the seed bank was released 95 days after the fire. A greater proportion of the seed bank emerged and established as seedlings 5 months after the fire in May than after a fire the previous February. Seedling emergence and mortality were broadly related to rainfall over this period. The relationship between various fire regimes and abundance in B. ericifolia populations is discussed.

Patterns of flowering, levels of fruit set and pollination limitation of fruit set were examined over four years in a group of four co-occurring Banksia species (Proteaceae) in south-eastern Australia

In coastal heath, 12 km north of Sydney, Banksia ericifolia set fruit after cross-pollination but not after controlled self-pollination. Animals removed nectar and pollen from inflorescences during the day but not overnight. Introduced honeybees (Apis mellifera) and native birds (Meliphagidae, Zosteropidae) visited flowers frequently during the day, carried pollen on their body surfaces and were likely to transfer pollen between plants. Native bees, nocturnal moths and ants also visited flowers but were too infrequent, did not move between plants or foraged inappropriately to be important pollinators. No mammals were seen visiting flowers and the small mammals we caught carried no pollen. Exclusion experiments confirmed that pollination occurred during the day and not at night, but the experiments were insufficient to determine the relative importance of honeybees versus birds as pollinators. Only about 3% of the flowers of Banksia ericifolia developed follicles under natural conditions. Follicle production in this species was limited by resources and not by pollinators, since multiple cross- pollinations of flowers did not increase follicle production above the natural rate.