The Old World climbing fern, Lygodium microphyllum, is an invasive weed in the Florida Everglades and the leaf roll galling mite, Floracarus perrepae, is a proposed biological control agent. Field studies were conducted for one to two years at sites in its native range in Australia, New Caledonia, and India to evaluate the effect of climate on F. perrepae. Monthly counts of the proportion of L. microphyllum subpinnae (leaflets) with leaf roll galls were used to measure the incidence of damage caused by F. perrepae. Between sites the most significant weather variable was rainfall 14 to 28 days prior to sampling, with higher levels having a depressive effect on the incidence of leaf rolls. Within sites the mean maximum temperature was the only significant weather variable, showing a decrease in the incidence of leaf rolls above 27°C, and it was predicted that no leaf rolls would form above 35°C. The weather parameters in Homestead, Florida for 2002 were within the range of those evaluated in the eight native range field sites. Thus, we do not predict that climate will prevent the establishment of this biological control agent for L. microphyllum in southern Florida.
Lygodium microphyllum (Cav.) R. Br. (Lygodiaceae, Pteridophyta), Old World climbing fern, is native to the wet tropics and subtropics of Africa, Australasia, Asia, and Oceania (Pemberton 1998), and is an aggressive invasive weed of moist habitats in southern Florida with the potential to spread into Central and South America (Pemberton & Ferriter 1998; Goolsby 2004). A biological control program was initiated in 1997 and surveys for potential agents were conducted in Australia and South Asia (Goolsby et al. 2003). The eriophyid mite, Floracarus perrepae Knihinicki and Boczek, was the most widely distributed with several geographically specific genotypes identified (Goolsby et al. 2003; Goolsby et al. 2004a). Throughout its native distribution in Australia and Asia, F. perrepae causes significant damage to L. microphyllum. Feeding by the adults and immatures causes formation of leaf roll galls, leading to necrosis and premature defoliation of L. microphyllum pinnae, and the gradual debilitation of the plant (Goolsby et al. 2003; Freeman et al. 2005; Ozman & Goolsby 2005). Based on its narrow host-range (Goolsby et al. in press), and significant impact on L. microphyllum (Goolsby et al. 2004b), F. perrepae was prioritized for evaluation as a biological control agent (Goolsby & Pemberton 2005).
Ozman & Goolsby (2005) documented the biology and seasonal phenology of F. perrepae in Southeast Queensland, Australia, and found that it was active year round, with populations peaking when temperatures were cool and soil moisture levels were highest. In these studies, the proportion of mite infested, or incidence of rolled subpinnae, was measured monthly at four locations over a two-year period. The incidence of leaf rolls was used as an indicator of the mite’s impact on the fern. Because the effect of weather was the most significant factor in the phenology of the mite in Australia, we replicated this study in India and New Caledonia to determine if the genotypes from these locations were affected in the same manner. The Indian and New Caledonian F. perrepae genotypes were known to have unique biological differences, in that they were co-adapted to the local genotype of the fern (Goolsby et al. unpublished data), but little was known about their possible adaptations to climate. Therefore, we used the incidence of the proportion of mite induced leaf rolls (or plant damage) to evaluate the seasonal phenology of F. perrepae in other parts of its native range that were climatically different to Southeast Queensland.
Materials and Methods
Study Sites
The initial baseline studies were conducted in Southeast Queensland (QLD) from February 2001 to March 2003 on Bribie Island; at Gallagher’s Point (27°01.17’S, 153°06.53’E) and McMahon Rd. (27°04.33’S, 153°10.55’E) and at Logan; Carbrook Creek (27°41.30’S, 153°16.23’E) and Lagoon Rd. (27°40.01’S, 153°16.03’E). Samples were collected monthly from February 2001 to February 2003. These sites are seasonally inundated with fresh water and L. microphyllum grows as a climbing vine on paper bark trees, Melaleuca quinquenervia (Cav.) St. Blake. The site at Carbrook Creek is different in that L. microphyllum grows under a closed canopy of M. quinquenervia in deep shade as compared to the other sites with open canopies. The climate in southeast Queensland is subtropical. Rain falls mainly in the summer months, which are hot and humid. Winters are cool and dry with an average yearly temperature of 20.6°C and average rainfall of 1393 and 1256 mm at Bribie Island and Logan, respectively.
Sites in India were located in southern Tamil Nadu at Thomaiyarpuram, Nagercoil (8°19.00’N, 77°25.70’ E) and in Kerala near Ithikkara River, Quilon (8°51.95’N, 76°41.79’E). Samples were collected monthly from July 2002 to August 2003. At the Nagercoil site L. microphyllum was growing under a canopy of coconut palms in deep shade. The Quilon site was exposed to full sun, with the fern growing in a ditch up a roadside embankment. The climate in southern India is monsoonal with heavy rainfall from April to August and hot, humid summers. The average yearly temperature and rainfall for Nagercoil and Quilon are 30.0 and 27.5°C, and 905 and 2932 mm, respectively. Sites in New Caledonia were located in Province Sud near Noumea at la Coulee (22°14.09’S, 166°34.75’E) and Yaté (22°06.36’S, 166°56.08’E). Samples were collected monthly from May 2002 to November 2003. At both sites L. microphyllum grows in partial shade with M. quinquenervia. The climate in New Caledonia is subtropical with rain evenly spaced throughout the year with an average yearly rainfall of 1106 mm. The average yearly temperature is 23.4°C.
We obtained the weather data from the closest available weather station for each location, daily maximum and minimum temperature, RH at 9 a.m. and 3 p.m. and rainfall. From these the daily soil moisture index (between 0 and 1 at soil saturation) was calculated based on a simple model developed by Fitzpatrick and Nix (1969). Yearly rainfall averages for India and New Caledonia were obtained from CLIMEX 2 (Sutherst et al. 2004) and the Queensland Department of Natural Resources for sites in Australia (Queensland 2004). For comparison of the native range to the proposed area of introduction for F. perrepae, we used weather data from 2002 for the Florida Automated Weather Network (Florida 2004), for Homestead, Florida, USA, which is located near Everglades National Park.
Sampling Methods
Lygodium microphyllum grows as a twining vine, and each shoot or rachis is a true leaf, consisting of pinnae (leaflets) and subpinnae (subleaflets). Vines were typically found climbing up the trunks of trees, reaching up to 10 m with many yellowish-green, fertile or sterile pinnae branching off the main stem. Each pinna consists of 6-12 paired subpinnae, which are the smallest leaf unit. Fertile subpinnae are fringed with lobes of sporangia, with sterile subpinnae having a smooth outer margin. Floracarus perrepae feed on and cause leaf roll galls on the fertile subpinnae, but they did not prefer this leaf form. Therefore, the field phenological studies were based on counts of sterile subpinnae. Each month 50 newly expanded sterile pinnae were collected at each site along a transect and returned to the laboratory for counting. At each site the numbers of infested and uninfested subpinnae were counted for each pinna. This count provided a measure of the incidence of infested subpinnae (leaf rolls), or total mite damage, at each location.
Statistical Analyses
Based on previous field research, it was determined that leaf rolls were the result of F. perrepae feeding activity from between 28 and 14 days before field collection (Goolsby et al. 2004; Ozman & Goolsby, in press). Therefore mean values of weather variables for this two-week period were used for the analyses. An analysis of variance was performed on the proportional incidence of leaf rolls to determine which of the weather variables accounted for the most variation between and within sites. Sites were assigned a value of 1 if the L. microphyllum grew under a canopy of full shade, with a value of 0 for sites with partial shade or full sun.
Results
The incidence of leaf rolls was observed monthly for periods between one and two years at eight sites, two in India, two in New Caledonia, and four in Australia. The mean value of the proportional incidence of leaf rolls at each site is shown in the Table 1, together with mean weather variables daily maximum and minimum temperature, rainfall, soil moisture, and shade.
An analysis of variance was performed on the proportion of leaf rolls to determine which of the abiotic variables accounted for the most variation between and within sites. The rainfall of the previous two-week period, averaged over each site, fitted as a quadratic equation explained 90% of the variation between the 8 sites (P < 0.001, linear regression coefficient (b1) = -0.00323, SE 0.00031 and quadratic regression coefficient (b2) = 0.0000485, SE = 0.000009, F2,4 = 22.71) (Fig. 1). Shade explained a further 5% of the between site variation but only at (P < 0.07) with leaf rolls from shaded locations (0.048 SE 0.019, F1,4 = 3.22) higher than unshaded. The shape of the quadratic for rainfall indicated that rainfall decreased the incidence of leaf rolls steadily as it increased. However, after 72 mm of rainfall for a two-week period, the predicted proportional incidence of leaf rolls remained fairly constant as rainfall increased. Temperature, relative humidity, and soil moisture were not significant in the analysis between sites.
Within sites (after adjusting for mean site differences) the mean maximum temperature fitted as a quadratic explained the most variation in the incidence of leaf rolls (P < 0.001, b1 = 0.1598, SE 0.0532, b2 = -0.00319 SE = 0.00099, F2,138 = 7.95) (Fig. 2). From the shape of the fitted curve, maximum temperature appeared to decrease the incidence of leaf rolls as the temperature rose above 27°C, but below the prediction was fairly constant. At mean maximum of above 35°C no curls were predicted. There was no interaction between the variables and sites; hence, all sites had the same response to the weather variables. Relative humidity, rainfall, and soil moisture were not significant within sites.
The weather data from Homestead, FL was calculated for 12 hypothetical monthly sample dates in 2002. The mean rainfall of the 12 two-week totals was 90 mm with the mean maximum temperatures of the two-week sampling periods in a range between 23.1 and 32.4°C (Table 1).
Discussion
Floracarus perrepae is widely distributed in Australia and Asia across a range of tropical and subtropical climates (Goolsby et al. 2003). Within this distribution are several location specific genotypes that are adapted to their corresponding genotypes of their host, L. microphyllum (Goolsby et al. 2004a; Goolsby et al., unpublished data). From these studies, it appears that the specific genotypes represented in southern Australia, New Caledonia, and India responded similarly to the weather variables in their environments, since there was no interaction with sites in the within site analysis.
The most significant weather variable between sites was rainfall. Jeppson et al. (1975) reported similar effects from high rainfall during the rainy season of Asiatic monsoon climates, which caused major reductions in mite populations such as the tea spider mite, Oligonychus coffeae (Nietner). In our studies, high amounts of rainfall 28-14 days months prior to the sampling date had a negative impact on the incidence of leaf rolls, which is an indicator of the local population density. Persistent and heavy rainfall during this time period most likely dislodged the dispersing F. perrepae as they attempted to settle and induce leaf rolls. This may explain why the sites in Quilon, India and New Caledonia had the lowest mean proportional incidence of leaf rolls. Quilon receives a double rainy period during the monsoon, and heavy persistent rain is common. In addition, the Quilon site has high rainfall, is in the open, and exposed to the direct effect of rain. In contrast, the site in Nagercoil received less rain, the L. microphyllum stand was under a sheltering canopy of palms, and the incidence of leaf rolls galls was higher, within the range of values for the Australian sites. In New Caledonia, high rainfall during 2002-03, including that associated with cyclone Erica, (March 2003) appears to have been responsible for the lower incidence of leaf rolls. Additionally, both sites in New Caledonia are partially exposed with L. microphyllum growing under a broken canopy of M. quinquenervia, thus allowing for the full impact of rainfall.
The higher average proportional incidence of leaf rolls at the Australian field sites also may be due to the drought conditions, which have affected eastern Australia since 2001 (Queensland 2004). Floracarus perrepae at these field sites have been exposed to fewer episodes of high rainfall, and mortality of dispersing females may have been lower. Thus, the higher proportional incidence of leaf rolls. The coconut mite, Aceria guerreronis Keifer (Acari: Eriophyidae), appears to be affected similarly across its known distribution where it reaches higher densities in areas with drier climates (Moore & Howard 1996). The study by Ozman & Goolsby (2005) found populations of F. perrepae peaked when temperatures were cool and soil moisture levels were highest, which appears to contradict the findings of this study. However, the results from the continental comparisons of climatic effects further clarify the differences between the effects of soil moisture and rainfall on F. perrepae populations. Our results suggest that F. perrepae performs best when soil moisture is near saturation (promoting growth of L. microphyllum) and there are few periods of high intensity rainfall, which interfere with dispersal.
The mean rainfall for Homestead, Florida in 2002 was higher than that for any station measured in Australia. One could predict that populations of F. perrepae in the latter localities, would be similar to those in New Caledonia, where the mite caused considerable damage to L. microphyllum. The rainfall effect appears to have a limit as shown in Figure 1, in which the model predicts that effect will remain constant above 72 mm for a two-week period. This prediction is corroborated by the data from New Caledonia and India, which showed that high rainfall reduced the incidence of leaf rolls but the populations remained stable.
High temperatures also had a significant negative impact on F. perrepae populations within sites. The model predicts that the proportional incidence of leaf rolls will decrease as the mean maximum temperature rises above 27°C with no leaf roll formation above 35°C. None of the sites in the native range experienced mean maximum temperatures for a two-week period above 35°C. Similarly, in Homestead, Florida during 2002 the highest means for the summer months ranged between 31 and 32°C, respectively. High temperatures are important to consider for rearing and release of F. perrepae. Summer greenhouse temperatures can often reach mean maximum temperatures above 35°C. Optimum rearing environments should be maintained at a mean maximum temperature of 27°C. Field colonization of F. perrepae in Florida would be the most difficult in the summer months when temperatures and rainfall are the highest.
In conclusion, the genotypes of F. perrepae evaluated in this study from its native range in Australia, New Caledonia, and India were all negatively impacted by high rainfall and temperatures during leaf roll formation. Conversely, populations of F. perrepae reach their highest levels when temperatures are cool and episodes of high intensity rainfall are few. The climate of southern Florida falls within the parameters experienced in the native range of F. perrepae. Therefore, we do not predict that climate will prevent the establishment of this biological control agent for L. microphyllum.
Acknowledgments
The authors thank USDA-ARS, Office of International Research Programs, the Florida Department of Environmental Protection, and the South Florida Water Management District for financial support; Dr. Dave Walter (University of Queensland), Drs. Richard Greene and Ernest Delfosse (USDA-ARS, National Program Staff) for support and encouragement; Ryan Zonneveld (CSIRO Entomology) for collecting and processing the field samples, and John Lydon (USDA-ARS) and Marc Coombs (CSIRO Entomology) for reviewing the manuscript.