Passiflora tripartita var. mollissima
(banana passionfruit)

Historical confusion over the taxonomy of this and its close relatives makes earlier reports on both distribution and invasiveness hard to interpret. For example, the species previously known in Hawaii as Passiflora mollissima is now P. tarminiana (HEAR, 2012) and in New Zealand P. mollissima is now regarded as P. tripartita var. mollissima, although P. tripartita var. azuayensis also occurs there, as does  P. tarminiana, which was previously known as P. mixta (Heenan and Sykes 2003). However, P.mixta sens. strict. is still recognised in New Zealand, but is restricted to the Waitakere ranges, near Auckland.

Although there is no indication that any of these species are invasive in their native habitats, all are regarded as invasive in one or more countries. This invasiveness is exacerbated in all species by their climbing habit compounding the difficulty of control, and by the activities of birds and feral mammals in facilitating their spread. Observation indicates that spread is sufficiently rapid to effect the alteration of forest habitats in a short time, with blankets of Passiflora foliage covering trees, shrubs and the ground, thereby preventing natural regeneration and succession. Ecological processes are severely disrupted, and biodiversity threatened, because these native forests are home to many species that have been proposed for listing as threatened or endangered.

Passiflora is a large genus with species distributed widely throughout the world. The subgenus Tacsonia is restricted to the Andes of South America with about 50 species that grow above 2000 m altitude (Escobar, 1992). P. tripartita var. mollissima is native to the high elevation Andean regions of southern Colombia, Ecuador, Peru, Bolivia and Venezuela. There are some plantings in southern Australia, and the plant may have become naturalized in southeastern Australia, but this has yet to be confirmed. Plantings have also occurred in Madras in India and New Guinea. New Zealand seems to have a climate that is suitable for it and it is grown in small amounts. In Hawaii, it was introduced as an ornamental and for the fruit, but it has become a very aggressive forest-destroying weed on the Big Island and Kaua’i (Duarte and Paull, 2015).

Williams and Buxton (1995) stated, "All three Passiflora spp. (P. mollissima, P. pinnastistipula and P. mixta) of the Tacsonia group originating in the Andean highlands are now widespread throughout the world, but only P. mollissima [now known to be P. tarminiana (Hear 2012, Coppens d'Eeckenbrugge et al., 2001)] has become a serious weed in Hawaii."

The risk of accidental movement of P. tripartita var. mollissima seed or other plant parts is considered minimal. Deliberate transport by human activity as a crop or an ornamental is the most likely avenue of introduction. However, this species' invasiveness has become widely recognized, and because of this, the danger of its intentional introduction to areas where it could survive and spread is now considered to be lower. It is a federally listed noxious weed in Hawaii.

P. tripartita var. mollissima is a perennial woody vine native to the high elevation Andean regions of southern Colombia, Ecuador, Peru, Bolivia and Venezuela, where it is widely grown for its fruit (Pemberton, 1983; LaRosa, 1984, 1985; Markin et al., 1992; Coppens d'Eeckenbrugge, 2001). In this regard, it is known in these habitats only in cultivation and from escaped populations.

Coppens d'Eeckenbrugge (2001) reported that in its native environment it is not well adapted to grow under 2400 m, and that its fruit grow bigger at higher altitudes.

In New Zealand, Baars et al. (1998) suggested that sites with introduced lianes (Clematis vitalba, Lonicerajaponica and 'P. mollissima' – almost certainly P. tripartita var. mollissima) - were generally characterised by high soil pH, warmer and lower altitude sites, proximity to the coast, westerly aspects and early successional vegetation. P. mollissima also tended to be restricted to low forest canopies, an observation shared by Williams and Buxton (1995) who were probably also dealing with P. tripartita var. mollissima. Baars et al. (1998) also found that P. mollissima (once again probably P. tripartita var. mollissima) was not very specific in its choice of supporting species, being more common in disturbed or regenerating forest which often had a lower canopy height.

Baars and Kelly (1996) and Beavon (2007) found that light availability was important for the survival of P. mollissima (probably P. tripartita var. mollissima) seedlings, and that growth and reproduction were restricted to high-irradiance environments.

In habitats to which it has been introduced, P. tripartita var. mollissima colonizes native forests, where it spreads as dense layers that cover and shade trees, breaking them down by the weight of the vines. The vines also increase the surface area of supporting trees, rendering the trees vulnerable to breakage by high winds. Dense mats of foliage of this and related species on the ground smother and kill forest seedlings, preventing regeneration (Markin and Nagata, 1989). P. tripartita var. mollissima is not normally a weed of crops in its native habitats or where it has been introduced. Baars et al. (1998) in New Zealand found that P. mollissima was not very specific in its choice of supporting species, being more common in disturbed or regenerating forest which often had lower canopy height.

Genetics

The chromosome number of P. tripartita var. mollissima is 2n=18 (Wagner et al., 1990).

Physiology and Phenology

Williams and Buxton (1995), almost certainly working with P. tripartita var. mollissima, found that in New Zealand establishment is restricted to areas of high light levels and soil disturbance. P.tripartita var. mollissima vines in the forest canopy were found to originate from adjacent open areas up to 15 m away. Baars and Kelly (1996) and Beavon (2007) in New Zealand reported that for P. tripartita var. mollissima, light availability appears to be important for the survival of seedlings.

Reproductive Biology

Although P. tripartita var. mollissima does not appear to produce fruit apomictically (Beavon and Kelly, 2012), stems that are broken or cut can grow new roots if they contact moist soil (Williams and Buxton, 1995).

Self-incompatibility has not been observed in the Tacsonia group (Coppens d’Eeckenbrugge et al., 1997), and stigmatic movement does not occur during anthesis as is observed in the Passiflora group. Research in New Zealand on P. tripartita var. mollissima by Beavon (2007) and Beavon and Kelly (2012) investigated mating systems and reproductive traits. They and other authors pointed out that humming birds are responsible for pollination in the species’ native environments. According to Quintero (2009), honey bees are also important pollinizers helping to improve fruit size and quality. In New Zealand, visitors to flowers were almost exclusively introduced honey bees (Apis mellifera) and bumble bees (Bombus spp.). Beavon and Kelly (2012) compared fruit set, seed set and germination success between hand-selfed, hand-crossed, bagged and open flowers. Bagging flowers to exclude pollinators reduced fruit set from 18.0% to 3.0%. Fruit set in hand-selfed flowers was lower, at 17.5%, than in crossed flowers, at 29.5%. However, bees did not perform as well as either hand-selfing or hand-crossing, indicating significant pollen limitation. The flowers are large in relation to bees and bees can visit the nectar source in them without necessarily contacting either anthers or stigma. Clearly, however, pollination by bees is adequate to allow the species to flourish in New Zealand and elsewhere. The different pollination methods tested by Beavon and Kelly (2012) had no significant effect on fruit size, nor on the number of seeds per fruit, which averaged 108. Neither was there any consistent effect of pollination treatment on seed germination. In Hawaii, abundant fruit set is observed. This seems to be due to a mixture of spontaneous self-pollination and pollination by insects. The newly opened flowers have exposed stamens, favourable to cross-pollination by insects; if cross-pollination does not occur, each flower later pollinates itself through movement of the stigmas to touch the stamens (Duarte and Paull, 2015).

Williams and Buxton (1995) found that fresh seed germinated over 12 weeks at 25^oC under constant low light, but that 73% remained ungerminated after that time. LaRosa (1992), in her studies on germination with P. tarminiana, found broadly similar results. Beavon (2007) found that seeds cleaned of flesh germinated in slightly higher numbers than did uncleaned seed, but there were significant differences between germination in the field (much lower, at 9% of cleaned seed) and in a glasshouse (66% of cleaned seed).

In its native environment, P. tripartita var. mollissima is capable of active growth, and of flower and fruit production throughout the year with little seasonal variation. Fruits are attractive to birds, pigs and small mammals, which consume the seeds with the fruit and may pass the undamaged seeds through their digestive tracts (LaRosa, 1984; Williams and Buxton, 1995; Beavon, 2007). Fruit production under cultivation in Bolivia is as high as 1000 kg of fruit/ha/week, and 13,000 kg/ha in 3 months during peak harvest (Casanas-Arango et al., 1996), indicating the potential for production and spread in areas of its introduction.

Environmental Requirements

P. tripartita var. mollissima is a species of mesic and wet, cool (10-20°C), upper-elevation habitats. It is a cool climate crop with ideal average temperatures of 14-16°C, conditions found between 1800 and 3200 m in the Andes (Fischer et al., 2009), with an optimum of 2200-2400 m altitude (Bonnet, 1988; Schoeniger, n.d.). Coppens d'Eeckenbrugge (2001) reported that in its native environment it is not well adapted to grow under 2400 m, and that its fruit grow bigger at higher altitudes. It grows at 3400 m in Cusco, Peru (Missouri Botanical Garden, 2003) and has adapted well to altitudes of 1200-1800 m in Hawaii and New Zealand. In New Zealand, Baars et al. (1998) found that sites where it grew were generally characterised by high soil pH, warmer and lower altitude sites, nearness to coast, westerly aspects and early successional vegetation. The plant can stand light frosts (Munier, 1961) and temperatures of -5°C for a short time. At higher altitudes, fewer anthracnose problems are experienced (Campos, 1992).

The optimal rainfall is between 1500 and 2000 mm (Quintero, 2009) uniformly distributed, otherwise irrigation is necessary for continuous fruit production. Dry periods occur in most of the Peruvian and Bolivian Andes where rain seldom reaches these amounts and normally occurs during only 6 months of the year (Duarte and Paull, 2015).

Both Baars and Kelly (1996) and LaRosa (1992) provided evidence that growth and reproduction appear to be restricted to high-irradiance environments, with ideal growing conditions of 1200-1500 h of sunshine per year (Campos, 1992). Photoperiod does not seem to have an effect on flowering since it does flower for long periods of the year at different latitudes.

Wind is very damaging to this plant, breaking young shoots and causing flower drop (MAG-INCCA, 1991). Fruit can also be damaged by rubbing with other plant parts. Pollinating insects are also affected by wind.

Soil type and soil pH appear not to be critical to the growth of P. mollissima. Plants have been grown experimentally in soils of various consistencies, as well as in sand and vermiculite (Gardner, 1989). The plant needs a soil profile that is at least 50-60 cm deep, medium textured (loam to sandy loam) and rich in organic matter (MAG-INCCA, 1991). The soil should have a good water retention capacity since the vine does not stand long dry periods. The area should not be subjected to flooding and requires good drainage. Campos (1992) suggests an ideal pH of 5.5-6.5. Casierra-Posada et al. (2011) explored the effects of soil salinity on P. tripartita var. mollissima, having commented that current agricultural practices in Columbia are tending to increase soil salinity. They found that with increasing salinity, number of leaves, shoot length, specific leaf weight, leaf area and dry matter all decreased, concluding that banana passionfruit seedlings are moderately sensitive to salt stress.

To support a biological control programme in Hawaii, Pemberton (1983, 1989) explored the Andean regions of Peru, Ecuador and Colombia in search of natural enemies of P. mollissima (which then included both P. tripartita var, mollissima and P. tarminiana). Later, Causton and associates conducted similar explorations in Venezuela (Causton, 1993, 1997; Causton et al., 2000; Causton and Pena Rangel, 2002). A number of insects associated with this species were identified as a result of these efforts, with those showing promise for biocontrol receiving more intense research attention. Causton's work focused on the fly Dasiops caustonae, which feeds on bud anthers and causes premature flower abscission. In Ecuador and Colombia, Trujillo and Taniguchi (1984) noted a defoliating moth as being of particular interest as a potential biocontrol agent: Cyanotricha necyria Felder (Lepidoptera: Notodontidae) (Markin et al., 1989). A second potential agent, a bud and fruit-feeding moth, Pyrausta perelegans (Lepidoptera: Pyralidae) was also found (Markin and Nagata, 1990). Valero and Viana (1970) reported chlorotic spotting of P. mollissima caused by a species of the cicadellid insect Empoasca sp. in Colombia. Josia fluonia (Lepidoptera: Notodontidae), a foliage-feeding moth from Ecuador, was introduced into quarantine in Hawaii for testing as a potential biocontrol agent (Friesen et al., 1994; Hennessey, 1996). Likewise, an apparently new species of the fruit fly genus Zapriothrica (Diptera: Drosophilidae), a pest of high elevation Passiflora spp., was identified in Colombia as a potential biocontrol insect and was evaluated for this purpose (Casanas-Arango et al., 1996) (see Biological Control).

Other insect pests include leaf eaters Dione or Agraulis juno that are found in groups. Fruit flies (Anastrepha spp.) can be sporadically found. Foliage can be infested by mites (Tetranychus spp.) (Duarte and Paull, 2015). In humid and poorly drained situations, nematodes (Meloidogyne spp.) can be a problem.

Few virulent diseases are known to attack P. tripartita var. mollissima; however, the following diseases have been reported: a stem and leaf spot caused by Colletotrichum gloeosporioides (USDA, 1960) and C. passiflorae, which was originally reported from Hawaii as causing anthracnose on leaves and fruit (Stevens, 1925). The rust fungus Puccinia scleriae attacks Passiflora species, including P. tripartita var. mollissima (Gardner and Davis, 1982). The alternate hosts of Puccinia scleriae are Scleria species (nut-rushes). Trujillo et al. (1994, 2001) found several fungal pathogens in the native habitats of Passiflora tripartita var. mollissima. Only two significant pathogenic fungi, capable of defoliating the host, were found on P. tripartita var. mollissima: a powdery mildew, Phyllactinia sp., and Septoria passiflorae (Ponte et al., 1979). Other published records of natural enemies of 'P. mollissima' (i.e. P. tripartita var. mollissima) include those of Chacon and Rojas (1981, 1984) and Morales et al. (2000).

Morphologically similar members of subgenus Tacsonia in the native habitats of P. tripartita var. mollissima, having flowers with tubular calyces, include P. tarminiana ('curuba de castilla') and P. mixta ('tasco de monte') (Escobar, 1980; Holm-Nielsen et al., 1988, Coppens d’Eekenbrugge et al., 2001). These forms are distinguished in field observation by variations in floral proportions, (e.g., floral tube length), presence and positioning of nectaries, pendent habit of the flowers, and by flower colour. P. tripartita var. mollissima flowers are fully pendent and a true pink to magenta, whereas those of P. mixta, for example, are semipendent and can be light pink to dark red. Hybridization among cultivated forms may further complicate morphological distinctions. Coppens d’Eekenbrugge et al. (2001) provide a useful key for distinguishing between P. tarmimiana, P. tripartita and P. mixta, and Heenan and Sykes (2003) provide a comprehensive key to the Passiflora species of New Zealand detailing the distinguishing characteristics of P. tarminiana.

Due to the variable regulations around (de)registration of pesticides, your national list of registered pesticides or relevant authority should be consulted to determine which products are legally allowed for use in your country when considering chemical control. Pesticides should always be used in a lawful manner, consistent with the product's label.

Cultural Control

Grazing cattle have been observed to effectively control P. tarminiana in Hawaiian forests, and Williams and Buxton (1995), in New Zealand, only found seedlings of P. tripartita var. mollissima in areas protected from sheep and cattle grazing. Rooting of pigs may also exert a temporary control, although Williams and Buxton (1995) pointed out that feral pigs are uncommon near human habitation, and in any case, disturbance caused by rooting is known to be deleterious to forest health in the long term and is not to be recommended.

Mechanical Control

Hand pulling of seedlings and small plants can be effective in a limited area, but this is very labour intensive and only useful for small infestations. As P. tripartita var. mollissima is a liana, and therefore closely associated with the overlain vegetation, finding all the stems can be difficult and removing them may involve damage to the supporting trees and other native vegetation.

Chemical Control

Herbicides that give effective control of Passiflora species include glyphosate and metsulfuron. They can be applied by knapsack, brush gun or by the cut stump method. Both herbicides will also kill or damage any desirable plant tissue they contact so they must be applied very carefully. Probably the best method of control involves cutting the vines back as low as possible in winter or early spring, and then spraying the regrowth later in spring with herbicide. The cut vines should be left hanging in the tree to dry out before being removed to prevent them from regrowing if they contact the ground, and to prevent damage to the host tree.

For larger plants the cut stump treatment works well. This means tracing the vines back to the root and cutting them off as close to the root as possible before immediately treating the cut surface with a suitable herbicide. This can be done by applying undiluted herbicide with a paint brush; alternatively, gel formulations of either glyphosate or picloram are available in some countries.

Using chemical control for vines like Passiflora spp. can be very laborious in large infestations and is best restricted to small or isolated populations.

Biological Control

Biological control has been long considered the most practical, cost effective, and long-range approach to P.mollissima (i.e. P. tarminiana) invasion in Hawaii (Waage et al., 1981; Gardner and Davis, 1982; Pemberton, 1983; 1989; Markin, 1989; Causton, 1997; Causton et al., 2000; Causton and Pena Rangel, 2002). In the late 1970s and early 1980s, Gilbert and his associates at the University of Texas at Austin, USA, expressed optimism that one or more species of heliconid butterflies from South America could be used successfully as biocontrol agents for P. mollissima in Hawaii, and with sufficient host specificity that the edible passion fruit, P. edulis, would not be threatened (Waage et al., 1981).

Aside from the efforts in Hawaii against P. mollissima (i.e. P. tarminiana), biocontrol work in New Zealand is outlined in unpublished Landcare research reports (Fowler, 1999; Winks and Fowler, 2000; Fröhlich and Gianotti, 2001).The Septoria leaf spot fungus was trialled as a biocontrol agent in New Zealand but, despite being virulent on weedy species of Passiflora, it also infected the crop species P. edulis and was therefore not considered safe for release (Hayes, 2005).

In 1982, the Hawaii Department of Land and Natural Resources provided funding support for exploration of the native Andean habitats in Peru and Colombia for potential biocontrol agents for P. mollissima, and Pemberton (1983, 1989) encountered several insects associated with P. mollissima though the necessary studies were not completed. Later, the leaf-feeding larvae of the moth Cyanotricha necyria were found to be effective defoliators (Lugo-Pena and Sanchez, 1974; Posada et al., 1976). Casanas-Arango et al. (1990) considered it to be the most important insect pest of P. mollissima in the Andean region (Ecuador-Colombian border) and to be readily available for collection throughout the year. C. necyria was the first insect introduced to quarantine in Hawaii in sufficient numbers to determine its biocontrol potential yielding positive results (Markin and Nagata, 1989). However, upon release into the field, C. necryia did not survive.

Likewise, Pyrausta perelegans, a bud and fruit-feeding moth from P. mollissima’s native habitats, yielded positive indications when tested in quarantine in Hawaii (Markin and Nagata, 1990; Casanas-Arango, 1996). Although P.perelegans became established in P. mollissima-infested forests on the islands of Maui and Hawaii, populations remain low and their impact on the weed is negligible. The reasons for the failure of these two species was thought to be due to predation by Lepidoptera and parasites already in the environment, released in Hawaii during previous years when indiscriminate release in attempts at biocontrol was practiced (Campbell et al., 1993; Casanas-Arango, 1996).

Other biocontrol agents have been tested for host specificity and found wanting for various reasons, including their likely impact on P. edulis. These include a species of Zapriothrica (Casanas-Arango, 1996) and the moth Josia fluonia from Ecuador (Friesen et al., 1994; Hennessey, 1996), as well as other species of flies.

The vascular wilt-causing fungus Fusarium oxysporum f.sp. passiflorae was reported to cause destruction in commercial plantings of the edible passion fruit P. edulis f. edulis in Australia in the 1950s (McKnight, 1951; Purss, 1954, 1958; Groszmann, 1958; Anon., 1960; Inch, 1978) and P. mollissima was also found to be susceptible (Gardner, 1989). The perceived threat to the passion fruit industry has prevented release the fungus into Hawaiian forests. Most of the pathogenic fungi found by Trujillo and associates in the native Andean habitats were on P. tripartita var. mollissima. Trujillo et al. (2001) did find two fungal defoliators on P.mollissima in Colombia that they considered sufficiently virulent to be potential biocontrol agents: a powdery mildew, Phyllactinia sp., and a leaf spot, Septoria passiflorae. The powdery mildew proved difficult to transport to Hawaii and its release was never accomplished. On the other hand, following host range testing, S.passiflorae was successfully released in heavily P. mollissima-infested forests at several sites on the islands of Maui and Hawaii (Trujillo et al., 1994, 2001) and has proven an effective defoliator of the target weed with no apparent damage to surrounding forest species or to other Passiflora spp. in Hawaii (Trujillo et al., 2001). Thus S. passiflorae continues to produce positive results and monitoring of its success is ongoing. Besides its application as a classical biocontrol agent, the possibility of developing S. passiflorae as a mycoherbicide, along with another classical biocontrol pathogen in Hawaii, Colletotrichum gloeosporioides f.sp. clidemiae, was also investigated (Norman and Trujillo, 1995). Again, results were favourable for S. passiflorae in comparison to C.gloeosporioides f. sp. clidemiae, although large-scale production of S. passiflorae as a mycoherbicide has not yet been undertaken.

Integrated Control

Taking into consideration the limited control possible with cattle or sheep grazing, mechanical means and herbicidal treatment, few control methods are available to integrate with biocontrol. This is especially true in forest preserves and national parks where manipulation by managers is kept to a minimum to preserve the natural setting as fully as possible.

30/11/2012 Updated by:

Ian Popay, consultant, New Zealand, with the support of Landcare Research.