Post-release support of host range predictions for two Lygodium microphyllum biological control agents

1 Introduction

The management of invasive species necessitates a careful and strategic approach that prioritizes the conservation of native communities. Classical biological control introduces host-specific agents into an invaded landscape with the intent of reestablishing the coevolved predator–prey relationship or herbivore–plant relationship. This approach aims to re-instill top-down consumer checks that were lost when the target species was introduced into its adventive range. In weed biological control, these reestablished relationships are expected to help suppress an invasive plant’s abundance in its adventive range through reciprocal self-sustaining population dynamics (McEvoy 2018; Müller-Schärer and Schaffner 2008). Successful classical biological control programs can lead to long-term, potentially widespread, cost-effective reductions of the invasive species (Hoddle et al. 2014; Müller-Schärer and Schaffner 2008; Winston et al. 2014).

The standards for release of biological control agents for invasive plants entail extensive pre-release impact assessments, focusing on host specificity by examining the range of plant species capable of supporting various life stages of the agent or serving as a food source (described in depth in USDA APHIS 2025; Van Driesche and Murray 2004). Prior to release, comprehensive physiological host range testing is conducted on target agents (Briese 2005; USDA APHIS 2025). This testing assesses the range of hosts that are physiologically compatible with the agent and evaluates the risk of direct nontarget impacts. One criticism directed towards classical biological control is the lack of post-release monitoring for nontarget effects (Carson et al. 2008). Many studies have found that the realized ecological host range of a herbivorous biological control agent is more constrained than the physiological host range because of limitations found in natural field conditions (Bowers et al. 2022; Hinz et al. 2014; Lake et al. 2015; Van Klinken 1999). Although post-release impact studies are not extensively practiced, in those examples where a field host-specificity examination was conducted, agents were found to perform as predicted from the quarantine host range tests (Bowers et al. 2022; Center et al. 2007; Lake et al. 2015; Morin et al. 2009; Pratt et al. 2013).

Less than 1 % of weed biological controls worldwide have produced adverse effects on nontarget plants (Hinz et al. 2019; Pemberton 2000; Suckling and Sforza 2014), though in New Zealand (where systematic post-release surveys are more common) 24 % of biological control agents have attacked nontarget plants (Paynter et al. 2004). Overall, nontarget attacks were predicted in host range testing (Hinz et al. 2019; Pemberton 2000) and in New Zealand the damage to nontarget plants was limited (Paynter et al. 2004). Although host testing has shown agents to be reliable in predicting field host ranges, post-release impact studies should be standard practice because they are an important check on whether biological control agents are performing as intended in the adventive range (Hinz et al. 2014; Paynter et al. 2004). Furthermore, post-release studies offer the opportunity to gauge effectiveness, assess agent performance relative to the numbers released, and identify unforeseen challenges. These challenges could include predation on agents, interactions with other biological control organisms, or susceptibility to diseases that were not apparent during laboratory testing. Post-release studies are opportunities for scientists to update predictions on non-target attacks, refine their approaches, proactively address potential issues, and ultimately improve the success rate of biological control initiatives while minimizing ecological risks (Paynter et al. 2018).

Lygodium microphyllum (Cav.) R. Br. (Lygodiaceae) is an invasive fern native to Australia, Southeast Asia, and Eastern Africa (Brandt and Black 2001; Goolsby et al. 2004, 2005) that has spread throughout Florida’s peninsula (EDDMaps 2024). This invasive fern encroaches into degraded habitats and ecologically sensitive environments, leading to further habitat degradation, diminished functionality, and reduced biodiversity (Boughton and Pemberton 2009; David and Lake 2020; Pemberton and Ferriter 1998; Volin et al. 2004). Contributing to its invasive abilities L. microphyllum exhibits high reproductive capacity and quick sexual development (Lott et al. 2003). Sori densely populate the underside of the fertile pinnules, with the potential to produce approximately 15,000 spores per cm² (Volin et al. 2004). The destructive and fast-growing nature of L. microphyllum prompted the initiation of a classical biological control program in 1997 (Goolsby et al. 2004).

Two L. microphyllum biological control agents have successfully established in southern Florida: the mite Floracarus perrepae Knihinicki & Boczek (Acariformes: Prostigmata: Eriophyidae) and the moth Neomusotima conspurcatalis Warren (Lepidoptera: Crambidae), with a third biocontrol agent, the moth Austromusotima camptozonale Hampson (Lepidoptera: Crambidae) not establishing (Boughton and Pemberton 2008). The eriophyid mite forms galls on leaflets and apical meristems of L. microphyllum, thereby substantially reducing the growth rate (David and Lake 2020). Additionally, galling from F. perrepae can cause leaf necrosis, reduced biomass, and reduced climbing ability (Goolsby et al. 2004). N. conspurcatalis may defoliate entire L. microphyllum thickets during population outbreaks but otherwise is found in isolated populations throughout the range (Boughton and Pemberton 2009). Larvae of N. conspurcatalis defoliate and skeletonize leaflets leaving behind clear epidermal cells, a process referred to as ‘windowing’. Eventually, the ‘windowed’ leaflets turn brown and die (Boughton and Pemberton 2009). Host range testing of F. perrepae on 32 species with 12 Pteridophyte and four Angiosperm families represented found the mite only developed on L. microphyllum, though there was limited oviposition and/or galling on three nontarget species, one being the Florida native Phlebodium aureum (L.) J. Sm. (Polypodiaceae) (Goolsby et al. 2005). The technique used in the host range testing increased the likelihood of a false positive because it focused on sporelings and not mature ferns. The United States Department of Agriculture Animal and Plant Health Inspection Service (USDA-APHIS) agreed with Goolsby et al. (2005) that feeding and development on nontarget species in field conditions was unlikely and issued a release permit in 2006. Based on information available about L. microphyllum genetics at the time, the F. perrepae mass rearing colony was obtained from individuals collected near Iron Range National Park in Queensland, Australia, which was thought to be the best match to the origin of the L. microphyllum populations in Florida (Goolsby et al. 2005). The moth N. conspurcatalis underwent host range testing on 47 plant species. These trials found that the moth was specific to the Lygodium genus, any oviposition and feeding on non-Lygodium species was limited to a few nontarget plants, with Dryopteris ludoviciana (Kunze) Small (Dryopteridaceae) and Osmunda regalis L. (Osmundaceae) having unusually high oviposition during one replicate of the multi-choice study (Boughton et al. 2009). In 2007 USDA-APHIS issued a release permit for N. conspurcatalis (Boughton et al. 2009).

The first releases of F. perrepae and N. conspurcatalis were in 2008 at Jonathan Dickinson State Park, Florida (Boughton and Pemberton 2009, 2011). From 2008 to 2010, small scale field releases of F. perrepae produced limited establishment but low population abundance (Boughton and Pemberton 2011; Lake et al. 2014). In contrast, N. conspurcatalis readily established persistent populations from the initial releases but was not able to survive two consecutive years of cold snaps in more northern release sites (Boughton et al. 2012; Smith et al. 2014). A mass rearing and release program for both biological control agents was initiated in 2014 as part of the Comprehensive Everglades Restoration Plan (David and Lake 2020; Lake et al. 2021). This program is ongoing and currently a capstone establishment survey is being conducted. At previous release sites included in the survey F. perrepae has established at 66.0 % of the sites and N. conspurcatalis at 81.0 % of the sites (unpublished data).

The main objective of this study was to conduct post-release validation of host range predictions for the two established biological control agents on L. microphyllum, F. perrepae and N. conspurcatalis, at four sites in Florida: Flatford Swamp, Strazzulla, Hungryland, and Weston (Figure 1A). This will help to validate the safety of these agents. Additionally, the present study assessed the presence of these agents in the four field sites. This helped document their ability to establish at sites within the host species’ adventive range.

Figure 1:

Survey site locations (A) for the field host specificity survey and (B) biological control agent monitoring numbers for the three east coast sites in Florida, Weston (WS, 26.09039°N, 80.412725°W), Strazzulla (SZ, 26.53043°N, 80.22859°W), Hungryland (HL, 27.000749 °N, 80.26848 °W), and Flatford Swamp (FF, 27.38112°N, 82.14023°W). The average number of Neomusotima conspurcatalis larvae was the mean of the counts for the two transects with the standard deviation illustrated via the vertical bars. Monitoring agent numbers for the west coast site were not obtained during the study.

2 Methods

3 Results

Despite seasonal examination over the course of one year, neither biological control agents (F. perrepae and N. conspurcatalis) themselves nor indications of their presence (i.e., oviposition, feeding, galling) were observed on non-target species during our study (Table 2).

The results of the presence/absence data show that F. perrepae was present at the three east coast sites but failed to establish at Flatford Swamp on the west coast (Table 1). Although galling was observed at the Flatford Swamp site, only two galls were observed during a single visit. On the east coast there was a general trend of galling observed outside the hottest months of the year, except for the Weston site where galling was observed on every visit. The presence/absence study indicated that N. conspurcatalis was present at all sites, although not consistently through all visits (Table 1). At the Flatford site N. conspurcatalis was present only in the beginning of 2022 and 2023 and absent the rest of the visits. At the Strazzulla site N. conspurcatalis was present every visit except the early 2022 visit because the site had received a herbicide treatment and there was not any living L. microphyllum available. The only time N. conspurcatalis was present at all sites (except Strazzulla because of the herbicide treatment) was the beginning of 2022, otherwise presence at sites was very variable.

The Weston site demonstrated seasonal fluctuations, as did Strazzulla, although the data were limited by unexpected herbicide treatments. Weston had the highest percentage of sub-pinnae galled in the fall of 2021 (75.0 %) and the spring of 2022 (66.4 %) with drops in the percentage of sub-pinnae galled in the colder winter months (14.2 %) and the hotter summer months (52.5 %). Strazzulla had low galling in the summer (2.78 %) but there was a substantial increase in the percentage of sub-pinnae galled in the fall (50.0 %). However, the Hungryland site failed to demonstrate the same seasonality in F. perrepae population abundance, peaking instead during summer (2021 42.4 %, 2022 48.1 %) (Figure 1B). N. conspurcatalis data were more sporadic and failed to display consistent seasonal patterns across sites (Figure 1B).

4 Discussion

Fundamental host range testing in a laboratory cannot, for logistical reasons, duplicate all factors that would successfully predict the behavior of a biological control agent in real world conditions. This limitation often leads to an overestimation of the agent’s potential host range (Haye et al. 2005; Hinz et al. 2014; Nechols et al. 1992). Post-release monitoring surveys, such as presented here, are useful for documenting these overestimations (Hinz et al. 2014). Our surveys confirmed this premise for N. conspurcatalis and F. perrepae in that despite some minimal damage or oviposition being observed on nontarget ferns during laboratory host trials (Boughton et al. 2009; Goolsby et al. 2005), none was observed in the field (Table 2). In fact, nontarget effects from these biological control agents have never been present at any L. microphyllum site since the inception of the mass rearing project (unpublished data).

Conducting surveys at different times throughout the year facilitated observations on agents during periods when species may exhibit varying degrees of palatability. Plants can emit different metabolic process signals throughout the year due to seasonal cues, indicating time-dependent processes (Gendron et al. 2021; Horvath et al. 2003; Wang et al. 2024). These processes can be complicated by different relative availabilities of target and nontarget plants. The presence-absence study on ferns adjacent to L. microphyllum revealed that N. conspurcatalis was present at all locations, albeit not consistently through time (Table 1). Predicting the establishment and timing of N. conspurcatalis population outbreaks has proven to be challenging (Boughton and Pemberton 2009; Jones and Lake 2020; Smith et al. 2014). Therefore, inconsistently finding N. conspurcatalis at our sites is congruent with other studies. F. perrepae was present at the three east coast sites but failed to establish at Flatford Swamp on the west coast (Table 1). Population monitoring showed the mites were extremely abundant on surveyed L. microphyllum and so could potentially have transferred onto nearby non-target fern species, but did not (Figure 1B). Previous research has suggested that F. perrepae populations are seasonal, with peak abundances in the fall and spring (David et al. 2019). This seasonal variation is likely influenced by climatic factors such as temperature and humidity which have been shown to affect F. perrepae (Goolsby et al. 2005; Muthuraj and Jesudasan 2011). Our findings partially support these contentions (Figure 1B). The Weston site, for example, demonstrated the expected seasonal fluctuations, as did Strazzulla although in the latter site data were limited by the unexpected herbicide treatment. The Hungryland site, however, failed to demonstrate this same seasonality in F. perrepae population abundance, peaking instead during summer (Figure 1B). N. conspurcatalis data were more sporadic and failed to display consistent seasonal patterns across sites (Figure 1B). As with F. perrepae, temperature extremes adversely affect establishment and limit N. conspurcatalis population numbers (Boughton et al. 2009). Most importantly, monitoring of the agents demonstrated that they had established self-perpetuating populations at the east coast sites.

The realized host range of agents can increase when populations of an agent increase relative to the host, which raises the chances of spillover effects (Ainsworth 2003; Paynter et al. 2020). Monitoring of L. microphyllum biological control agent populations indicated that both agents were present in large numbers, at least periodically, at the Weston and Hungryland sites thereby increasing the possibility of spillover. The two other sites had unique factors that could have contributed to spillover events if the agents were capable of impacting nontarget ferns. Strazzulla, the first of these two sites, was treated with herbicide in late 2021. This treatment theoretically increased the risk of spillover because herbicide treated sites have reduced host plant densities and quality, which may encourage agents to seek new plants for feeding and/or oviposition (Ainsworth 2003; Carruthers et al. 2023; Haag and Habeck 1991). Similarly, nearly 50,000 moths and over 2 million mites were released at the Flatford Swamp site during the course of this study, yet no spillover onto nontarget ferns was observed. To put this in perspective 2 million mites is more mites than were released for each of the eight first years of the mass rearing and release program (Aquino-Thomas et al. 2025). As agents are released at a site the chances of the agents finding suitable unoccupied habitat on the target species decreases and the likelihood of nontarget effects increases as the agents spread from the release locations looking for suitable habitat (Ainsworth 2003; Menéndez et al. 2002; Paynter et al. 2020). This is especially noteworthy because L. microphyllum in parts of the west coast of Florida, including Flatford Swamp, appears to be resistant to galling (Boughton and Pemberton 2011), a trait which could have caused F. perrepae to move off the target fern and onto nontarget species, but again it did not.

This study supports previous determinations on host specificity of the agents on L. microphyllum, as predicted by laboratory host range testing, with no evidence of nontarget impacts. The absence of spillover (oviposition, feeding, or galling) on nontarget ferns, even in herbicide treated and high-release sites, reinforces how safe these agents are for release. While the presence-absence data and population monitoring revealed natural fluctuations and challenges in predicting agent establishment and timing, the overall findings support the ability of these biological control agents to persist in the environment. This study highlights how field-based host specificity is likely to be more limited than laboratory host specificity trials indicated (Bowers et al. 2022; Hinz et al. 2014; Lake et al. 2015, Van Klinken 1999). Future biological control research should incorporate field-based post release monitoring to complement laboratory assessments to confirm their ecological safety and persistence.