1. Introduction
The
Cactaceae is a large family of succulent plants that comprises more than 120 genera and 1500 species [
1]. Almost all
Cacti are native to the Americas but were introduced around the globe either deliberately or accidentally [
2]. Many are cultivated for ornamental purposes, food and various industrial uses [
2,
3], but a large number have also become major weeds in many countries. These plants have many unique features (e.g., Crassulacean Acid Metabolism (CAM)) that allows them to establish and thrive in harsh and dry environments [
2,
4], making them a major threat to rangeland environments that they invade.
The majority of soil types and climatic regions in Australia are favourable for cacti growth and although the continent has no native cacti species, many exotic species have become naturalised following their deliberate introduction, mainly for ornamental purposes [
3,
5,
6,
7,
8]. Naturalised cacti populations can form dense, impenetrable thickets that limit access to grazing activities and reduce habitat quality. Furthermore, the spiny stems pose a significant health risk to humans, livestock and wildlife. As a result, 27 cacti species are listed as Weeds of National Significance (WONS) in Australia, all of which belong to the Opuntioideae sub-family and the genus
Opuntia and
Cylindropuntia [
3]. However, several
Cacti species from other genera such as
Epiphyllum,
Harrisia and
Cereus are also emerging or major weeds in different parts of the country [
3,
9].
Cereus uruguayanus Ritt. ex Kiesl. of the Cactoideae sub-family is a large, columnar cactus (
Figure 1a) that generally forms a spiny, multi-stemmed candelabra [
9]. Its taxonomy has been confusing with it being called several other synonyms in the past, including
C. peruvianus and
C. hildmannianus [
9,
10]. Common names used in Australia include Willows cactus, apple cactus, night-blooming cereus and torch cactus. The species originates from South America and is believed to have been introduced to Australia for ornamental purposes. It is a common plant in Queensland gardens because of its interesting shape, large white flowers and edible fruits [
11].
Cereus uruguayanus is not a WONS nor a declared weed in any state or territory of Australia, but the presence of large and multiple infestations in the Central Highlands region of Queensland has led to it being declared a priority pest species under local government legislation. Large infestations also occur in southern inland Queensland at several locations and many smaller infestations are common across southern/central Queensland [
11], as well as a limited number of locations in New South Wales [
12]. The species is most prevalent in mixed Eucalypt–Brigalow woodlands on light clay soils but also occurs in areas of cleared improved pasture and on a range of soil types.
Endozoochory (i.e., seed dispersal via ingestion by vertebrates) appears to be the primary dispersal mechanism of
C. uruguayanus (
Figure 1b) [
11] and was attributed to the occurrence of isolated plants up to one kilometre away from a naturalised population [
9]. Similarly, the invasiveness of the related cacti species
C. jamacaru in South Africa was largely attributed to this dispersal mechanism [
13]. Control options that promptly arrest flowering and fruiting are therefore desirable.
A potential
C. uruguayanus biological control agent, Harrisia cactus mealybug
Hypogeococcus festerianus (Hemiptera: Pseudococcidae), is already present in Australia and has been established on
C. uruguayanus at several locations in Queensland [
11]. Anecdotally, it appears to be causing some damage to
C. uruguayanus (Craig Hunter, personal communication), whilst in South Africa it provided effective control of
C. jamacaru [
13]. However, ten years or more was required for the agent to make an observable impact on
C. jamacaru population structure. Effective control also requires manual redistribution of the mealybug to uninfected weed populations due to its limited natural dispersal abilities [
14].
Chemical control offers a more immediate solution for treating high-priority infestations, particularly those in the early stages of establishment. A range of techniques and herbicides were investigated in Australia for control of Opuntioid cacti, but not
C. uruguayanus. As a result, there are several registered herbicides for use on Opuntioid cacti in Australia [
3], but their efficacy against
C. uruguayanus is largely unknown. In South Africa, foliar spraying of small plants or stem injection of large plants with monosodium methylarsonate (MSMA) is used for chemical control of the closely related species
C. peruvianus [
14] and
C. jamacaru [
15]. Testing is needed to determine whether MSMA also provides effective control of
C. uruguayanus.
The primary aim of this study was to identify effective herbicides to control C. uruguayanus using a range of techniques. For all experiments, we tested the hypotheses that at least one of the herbicide/technique combinations used would (Alternate hypothesis) or would not (Null hypothesis) kill C. uruguayanus plants. An initial screening trial compared the control efficacy of MSMA with other herbicides registered for use on Opuntioid cacti in Australia. Application techniques were based on herbicide label recommendations and included basal bark and cut stump application, foliar spraying and stem injection. A second experiment investigated the efficacy of soil-applied residual herbicides, as anecdotal evidence from landholders suggested that C. uruguayanus was susceptible to tebuthiuron and hexazinone-based products. Promising treatments identified in the screening trial were the focus of a later, third experiment that compared cut stump application of glyphosate at varying rates with cut stump application of a triclopyr/picloram mixture.
2. Materials and Methods
2.1. Site Details
All experiments were undertaken near the township of Willows (23°44′S, 147°32′E), approximately 70 km west of Emerald, Queensland, Australia. Climatically, the area has a summer dominant rainfall (Based on Gemfields Township records) which averages 589 mm annually and ranges from 21.5 mm in July to 104.0 mm in January [
16]. In terms of prevailing temperatures, monthly mean minima (Based on Emerald airport records) range between 9.1 °C (July) and 22.3 °C (January) while the monthly mean maxima range between 23.4 °C (June) and 34.6 °C (January) [
17,
18].
This area falls within the Brigalow Belt of Queensland and is characterised by clay soils, deep depressions (gilgais) and dense woodland vegetation dominated by the tree Acacia harpophylla F.Muell. ex Benth. (commonly referred to as brigalow). Experiments 1 and 2 were undertaken adjacent to each other in a remnant patch where the tree layer comprised A. harpophylla and several Eucalyptus spp., the mid-story C. uruguayanus and Carissa ovata R.Br. and the understory was sparsely covered with Aristida spp., and occasionally Cenchrus ciliaris L. Experiment 3 was undertaken nearby in a cleared area that had been planted with C. ciliaris. It dominated the ground layer, with isolated regrowth of A. harpophylla and C. uruguayanus scattered throughout the area.
2.2. Experiment 1. Screening for Suitable Herbicides and Application Techniques
The screening experiment was established in May 2016 and employed a completely randomised design with two replicate plots per treatment. Twenty-one treatments involving twelve herbicides registered for control of various cacti species in Australia and four different application methods (
viz. basal bark, cut stump, foliar and stem injection) were selected to test their efficacy against
C. uruguayanus (
Table 1). Untreated plants served as a control treatment. In terms of modes of action (MoA), most herbicides belonged to Group 4 (Disruptors of plant cell growth; auxin mimics). The exceptions were glyphosate [Group 9: Inhibition of 5-enolpyruvyl shikimate-3 phosphate synthase (EPSP inhibition)], metsulfuron-methyl [Group 2: Inhibition of acetolactate synthase (ALS inhibitors), acetohydroxyacid synthase (AHAS)], amitrole (Group 34: Inhibition of lycopene cyclase), and MSMA (Group 0: Herbicides with unknown mode of action) [
19].
Each experimental unit comprised 20 plants (spaced at least 2 m apart), approximately half of which were classified as small (<2 m in height) and the other half as large (≥2 m in height). On average, small plants were 1.03 ± 0.05 (SE) m high with an average basal diameter of 4.12 + 0.20 (SE) cm and large plants were 3.43 m ± 0.17 (SE) m high with an average basal diameter of 10.08 + 0.49 (SE) cm. As a result of the heterogeneity of the site, individual plot areas varied (c.a. 100–200 m−2) to accommodate the required number of plants in each size class. A 2 m buffer zone was maintained between the plots.
All cut stump treatments (except aminopyralid/picloram) and basal bark treatments were applied in May 2016. Stem injection and foliar treatments were delayed until October 2016 when C. uruguayanus plants were in a healthier condition. At this time, we also included an aminopyralid/picloram cut stump treatment to see if this portable and easy to apply Gel (Vigilant™ 11 paste; Corteva Agriscience Australia Limited, Chatswood, NSW, Australia) had the potential for landholders to treat isolated plants when they came across them during their day-to-day work activities.
Basal bark treatments were applied at an average operating pressure of 70 kPa using an 8 L handheld pneumatic sprayer (Swissmex®; Croplands Australia, Dry Creek, SA, Australia) fitted with a 0.6 m wand and an adjustable full cone nozzle. Herbicide mixture was applied to the point of run-off to the full circumference of the basal 5 cm (“thinline” treatment) or 40 cm (“traditional” treatment) of each plant stem. Cut stump treatments involved cutting off the plants ~10 cm above ground level using a battery-powered saw. Herbicide mixture was applied within ~5 s to the cut surface of the stump using the same equipment as described for the basal bark treatments. Foliar treatments involved spraying the whole plant (average operating pressure of 175 kPa) to the point of run-off using a 15 L backpack sprayer (Swissmex®; Croplands Australia, Dry Creek, SA, Australia) fitted with an adjustable solid cone nozzle. Stem injection treatments used a cordless drill with a 9 mm bit to insert 2–3 cm deep holes on a 45° downward angle at 10 cm intervals (hole centre to hole centre) around the circumference of each plant, at a height of ~40 cm. An NJ Phillips tree injection gun® (NJ Phillips Pty. Ltd. Limited, Somersby, NSW, Australia) was then used to apply either 1 mL (amitrole/ammonium thiocyanate treatment) or 2 mL (all other treatments) of herbicide mixture into each hole.
Plants treated in May 2016 were assessed 6, 12, 17, 25, 31 and 42 months after treatment (MAT). Plants treated in October 2016 were assessed 1, 7, 13, 20, 26 and 38 MAT. At each assessment time, plant injury for basal bark, foliar and stem injection treatments was scored on a scale of 1 (alive) to 10 (dead), with each incremental increase representing a 10% increase in the proportion of dead plant material. From this, plant mortality (%) was calculated as the number of plants with an injury score of 10, expressed as a percentage of the total number of plants of that size class in the experimental unit. For cut stump treatments, separate injury scores and mortality rates were recorded for stumps and fallen stems. The same method as mentioned above was used for the fallen stems, but for the cut stump, the rating was confined to a 1 (representing alive) or a 10 (representing dead). At each assessment time, the presence/absence of flowers and fruits was also recorded.
2.3. Experiment 2. Evaluating Efficacy of Soil Applied Residual Herbicides
The residual herbicide experiment was initiated in November 2016 and employed a randomised complete block design comprising eight treatments and three replications. Treatments included soil application of tebuthiuron at 0.4, 0.8, 1.2 or 1.6 g a.i. m−1 plant height (applied as 2, 4, 6 or 8 g m−1 Scrubmaster®, 200 g kg−1 a.i.; FMC Crop Protection Australia, North Ryde, NSW, Australia) and hexazinone at 0.5 or 1.0 g a.i. m−1 plant height (applied as 2 or 4 mL m−1 Bobcat® SL, 250 g a.i. L−1; Adama Australia, St Leonards, NSW, Australia). Stem injection of hexazinone was also tested in this experiment as it was not permissible within the remnant vegetation site used for Experiment 1. Untreated plants served as a control treatment.
Each experimental unit consisted of clusters of 10 plants usually located within a 200 m−2 area, although occasionally slightly larger areas were needed due to the variable distribution of C. uraguayanus at the site. On average, plants were 1.35 ± 0.04 (SE) m high with an average basal diameter of 8.55 + 0.31 (SE) cm.
For soil-applied treatments, pre-weighed tebuthiuron granules were sprinkled evenly within a 10 cm radius of the base of plants, whereas the liquid formulation of hexazinone was injected ~5 cm below ground using an NJ Phillips tree injection gun® (NJ Phillips Pty. Ltd. Limited, Somersby, NSW, Australia) with a spear and brace attachment. Injections were applied within 10 cm of the base of plants and if more than one dose was required, they were evenly spaced around the base. Stem injection of hexazinone was conducted using the same technique described for Experiment 1 and resulted in a mean dosage of 0.53 ± 0.06 g a.i. m−1 (± s.e.). Plants were assessed 6, 11, 19, 25 and 36 MAT for plant injury score and plant mortality (as described in Experiment 1).
2.4. Experiment 3. Optimising Glyphosate Dose for Cut Stump Application
The cut stump experiment was initiated in November 2017 and employed a randomised complete block design, incorporating six treatments and three replications. Treatments included cut stump applications of glyphosate at rates of 45, 90, 180 or 360 g a.i. L−1 [Applied as 125, 250, 500 or 1000 mL L−1 Roundup® (360 g a.i. L−1 present as isopropylamine salt)]. They were compared for C. uraguayanus control efficacy with cut stump application of a diesel-based mixture of 4 g a.i. L−1 triclopyr and 2 g a.i. L−1 picloram applied as 17 mL L−1 Access® (240 g a.i. L−1 triclopyr, 120 g a.i. L−1 picloram). Cut stumps without herbicide application served as a control treatment.
Each experimental unit was 20 C. uraguayanus plants, with half classified as small (<2 m in height) and the other half as large (≥2 m in height). On average, small plants were 0.98 ± 0.03 (SE) m high with an average basal diameter of 3.5 + 0.14 (SE) cm and large plants were 3.51 m ± 0.10 (SE) m high with an average basal diameter of 10.1 + 0.34 (SE) cm. Individual plot area averaged approximately 100 m−2 but varied slightly to accommodate the required number of plants in each size class. A 2 m buffer zone was maintained between the plots.
All glyphosate treatments used water as a carrier and contained 2 mL L−1 Pulse® Penetrant (1020 g a.i. L−1 Polyether modified polysiloxane). Herbicides were applied using the same technique described for cut stump treatments in Experiment 1, with the exception that the cut surface of the fallen stem was also treated.
Plants were assessed 7, 13 and 25 MAT for plant injury score and plant mortality of both the stump and the fallen stem (as described in Experiment 1).
2.5. Statistical Analysis
All data analyses were conducted using either Minitab
®, Version 17.3.1 (Minitab Pty Ltd., Sydney, Australia) or R and the Agricolae package [
20,
21]. Data expressed as percentages (i.e., mortality) were arcsine transformed prior to analysis and later back-transformed for presentation in tabular and graphic format. For each evaluation time, data from Experiment 1 and 3 were subjected to two-way analysis of variance (ANOVA) using the split-plot design (i.e., herbicide as main plot and size as sub-plot) to test whether the size of plants influenced the efficacy of the applied herbicide treatments. Data from Experiment 2 were subjected to a one-way analysis of variance for each evaluation time. With implementation times for the different techniques used in Experiment 1 split into either May or October due to the condition of plants, separate statistical analysis was undertaken based on when the treatments were applied.
Using ANOVA the significance of the sources of variation were assessed using the Fisher-Snedecor test. If the studied variable (i.e., treatments) was found to have a p-value lower than the significance level of α = 0.05 it was deemed to be significant, and a Fisher’s protected least significant difference (LSD) test was then used to undertake pairwise comparisons of means of all possible treatment combinations. Those combinations were identified as significantly different if their p-value was lower than the significance level of α = 0.05. As the two-sided LSD test p-value is twice that of a one-sided test, the direction can be assessed on significant results.
4. Discussion
While there is a broad range of herbicides recommended for the control of invasive cactus species in Australia, using various techniques [
3], this study found that a smaller number were effective on
C. uruguayanus. Even those effective were often slow to cause high mortality (>90%) of
C. uruguayanus, particularly on larger plants which tended to take longer to die than smaller ones. Nevertheless, at least one herbicide demonstrated high efficacy using basal bark (triclopyr/picloram), cut stump (aminopyralid/metsulfuron-methyl, glyphosate, metsulfuron-methyl, triclopyr/picloram, triclopyr/picloram/aminopyralid), stem injection (glyphosate, MSMA, triclopyr/picloram/aminopyralid) and foliar techniques (aminopyralid/metsulfuron-methyl, MSMA, triclopyr, triclopyr/picloram/aminopyralid) due to their ability to kill both small and large plants. Ground application of residual herbicides was less conclusive and warrants further investigation.
Basal barking was one of the slowest acting techniques particularly on larger plants, which recorded minimal mortality (<20%) 17 MAT. However, using triclopyr/picloram high mortality (≥90%) was eventually achieved 25 and 31 MAT, for small and large plants respectively. This slow activity is partly due to the way the herbicide affected the plant. We observed that necrosis occurred initially where the herbicide was directly applied (i.e., bottom 40 cm of the stem). The remaining upper portion of the plant would then slowly die over time while still upright, although the stems of several larger plants fell over and died lying on the ground. Differences in response of small and large woody plants to basal bark treatments are not uncommon, with smaller plants generally more susceptible [
22,
23,
24].
Both the traditional and thinline basal bark methods are used to control invasive woody weeds [
22,
23,
24,
25,
26]. In the current study, they provided similar efficacy and either could be used depending on the preference of the operator.
Leucaena leucocephala (Lam.) de Wit is another species that was also found to be equally susceptible to both basal bark techniques [
25].
Compared to basal barking, cut stump applications were much quicker to cause high mortality of treated stumps for all herbicides that were tested. This is consistent with what was achieved for a range of woody weeds using this technique, provided plants are cut off close to ground level and the herbicide is applied immediately onto the cut surface to allow rapid absorption [
26], as occurred in the current study. Experiment 3 demonstrated that the application of herbicide onto the cut stump is essential for high mortality. In the absence of herbicide, mortality of the cut stumps averaged only 31% and 58% for small and large plants, respectively. Even lower mortality was reported for woody weeds such as
Ligustrum sinense Lour. (14% mortality) [
27] and
Ailanthus altissima (Mill.) Swingle (21% mortality) [
28] if herbicide was not applied to the cut stump. In the current study, the height that plants were cut off was
c.a 10 cm aboveground which caused high mortality after herbicide application. At greater cut heights, plant mortality could be expected to decline as was reported for several woody weeds [
26,
29]. For example, mortality of
Calotropis procera (Aiton) W.T. Aiton after application of 2,4-D butyl ester averaged 67% when plants were cut 5 cm above the ground and 0% when the plants were cut 20 cm above ground level [
26].
While the stumps of plants treated using the cut stump method died relatively quickly, the fallen stem sections took much longer to die. In some instances, fallen stems reattached to the ground and new stems shot along the length of the original stem. This was a rare occurrence following cut stump applications in May 2016, but more common when undertaken in October 2016. This may be associated with the duration of dry conditions following treatment, with prolonged dry periods potentially facilitating higher mortality. Plants treated in May were exposed to the whole dry season period (although it was wetter than average), while those treated in October experienced a much shorter dry period before the start of the wet season which typically commences by November in sub-tropical/tropical environments. Spraying of the cut section of fallen stems to increase mortality proved successful for small plants but not larger ones. Despite the limited regrowth from fallen stem sections, landholder experience associated with other tall-growing opuntia cactus often makes them hesitant to undertake cut stump applications. A big difference however is that we observed that
C. uruguayanus maintains its structure when it falls onto the ground, whereas cladodes and stem sections readily break off on many other cactus species. It is these cladodes/sections that can result in the formation of a new plant [
3].
Foliar spraying demonstrated the greatest variability of all the techniques tested. Across the eight foliar herbicide treatments, the rate at which plants died and overall mortality varied markedly. Some herbicides were ineffective on small and large plants, others only controlled small plants, while a limited few caused high mortality irrespective of the size of plants. MSMA was the fastest-acting herbicide and killed
C. uruguayanus plants more quickly than all other treatments. It is used for the control of
C. peruvianus [
30] and
C. jamacaru [
31] in South Africa and is particularly effective on smaller plants. It is also recommended for several Opuntioid cacti in Australia, but due to its poison status, strict safety guidelines are prescribed to minimise risks to the operator. In the current study, three slower-acting herbicides eventually caused very high mortality of both small and large
C. uruguayanus plants. They were two formulations of triclopyr/picloram/aminopyralid (Grazon™ Extra and Tordon™ RegrowthMaster) and triclopyr. They are all considered less hazardous than MSMA and are commonly used for spraying a range of other invasive woody weeds and several other cactus species [
3]. While they were capable of causing high mortality of large plants, from an application perspective this technique would be most suitable for treating smaller plants (i.e., ≤2 m) with larger plants controlled using other options, such as stem injection.
Stem injection was highly effective at killing both small and large plants of
C. uruguayanus, with glyphosate performing as well as MSMA despite being slower acting. These herbicides are recommended for control of other cactus species [
3,
30] using the stem injection technique given their succulent nature and the ability of systemic herbicides to translocate rapidly. Minimisation of damage to non-target species and low costs are also considered advantages of the stem injection technique for cactus control [
30,
31]. The broad spectrum triclopyr/picloram/aminopyralid (Tordon™ RegrowthMaster) also proved effective and would be an appropriate option in areas where there are multiple woody weeds growing amongst
C. uruguayanus. Hexazinone was also highly effective when stem injected, but its use would be restricted to more open areas to minimise non-target impacts.
Ground application of two residual herbicides failed to cause high mortality of C. uruguayanus, but further investigation is warranted. While hexazinone performed poorly, tebuthiuron demonstrated a linear rate response, so testing higher rates may result in higher mortality of C. uruguayanus and should be explored. Furthermore, no non-target damage was observed in the stem injection treatment, but despite being injected into the soil dead C. ciliaris tussocks were observed around all ground applied hexazinone treatments. It was particularly evident on the lower slope of treated plants where in some instances dead C. ciliaris plants were up to 3 m away.
Whilst this study has identified several effective herbicides and application techniques for the control of
C. uruguayanus, further research is warranted. In particular, future studies should focus on identifying the most cost-effective control options for infestations of differing sizes and densities. This should include exploration of other potential techniques (i.e., mechanical, biological) as some herbicides can cause adverse effects on the environment and they may not be applicable in all situations, such as environmentally sensitive areas. This is particularly pertinent for residual herbicides such as tebuthiuron and hexazinone that can remain in the soil for extended periods [
32,
33], but even herbicides that are thought to be relatively safe to use (e.g., glyphosate) could be problematic in some situations [
34]. Irrespective, a single technique is rarely effective for control of a particular weed and an integrated approach is usually needed to deal with not only the original plants but also the subsequent regrowth that could continue to appear whilst there is a residual seed bank, or if the site continues to be re-infested from external sources (such as a nearby infestation) [
35]. An enhanced understanding of the ecology and population dynamics of
C. uruguayanus would also be advantageous and provide insights into how this plant is likely to respond to imposed treatments over time.