News and Updates

Building the algorithm for rubber vine

September 2023

The rubber vine program in the Kimberley the KRBA is partnering is looking at improving its surveillance methods for the weed.  Up until now the West Kimberley site used the method of taking images from the air across the 18,000ha site and searching the images by means of computer looking for the white flowers of adult plants showing up against the green foliage background.  On the East Kimberley site surveillance of the 30,000ha site was also carried out from the air but using human observers.  Both had some advantages but also a number of weaknesses.  In the case of the West Kimberley method only mature plants that were flowering on the day could be located. The East Kimberley had the disadvantage of the observers missing plants as they tired over the day.

The new option is an upgrade of the West Kimberley method using very high resolution cameras that will allow small leaf and stem structures of different plants to be accurately identified from an image, then teach a machine to pick up on those key identification points humans use and make a decision of yes or no on a plant.  The critical issue is how much you can magnify the image before pixelation turns the image to mush.

In a field trial 11,000 images were collected by drone of which 2,000 were searched by eye to locate plants.  This was based on a ground search of the site that located numerous plants to guide the image search.  The  information will all go to develop the algorithm to enable rubber vine plants as well as the flowers to be found and the remaining 9,000 images will then be searched using the algorithm and another on site from the ground to test the process.  If both searches align it will indicate some form of initial success.

From there hopefully the new method will be used successfully to cover both sites.


Mesquite work south of Derby

September 2023

The KRBA surveyed and controlled mesquite on just over 4,250 hectares of land just south of Derby in April this year.  The work had been delayed from January due to the extreme flooding at that time. 

In all very few plants were located, just 43 in total.  The biocontrol that was found in the area in 2022 was still active, though nowhere near as active as it was in that July.  It’s suspected that the moth that desiccates the flowers and leaves of the mesquite plant would have periods of high and low activity according to the climatic conditions with the wet season being a low point.

The interesting thing about this site currently is the lack of small seedlings.  Mesquite does not compete well with thick grasses and the fact that a biocontrol is putting pressure on the plant’s seed production sites could be reducing that new recruitment. 

Just about all plants were over two metres in height with some in the more heavily timbered area up to six metres high.  The height of those plants does expose one weakness of aerial control in heavy timber, that being you have to wait until the target plant is big enough to have grown out from underneath a larger native tree that is sheltering it.  Flying low and  looking across as much as down does help but plants do remain invisible often for a number of years before being noticed.  The other side of the coin is that you could never cover that amount of ground in just a couple of days on foot or vehicle.  It would take weeks of labour gridding at distances as little as 20 metres to do the same job.

An interesting development coming from the flood was that it was noted that a number of sites that did have mesquite actively growing on them had disappeared.  In all, just over 400 hectares of vegetated river flat had been washed away, along with it the mesquite growing on it.  See the image below showing 2022 river banks and the red track line showing the current banks.  Even the pilot observed that one of his fishing spots was no longer in existence, its remnants was now 500 metres off shore.

Where these mesquite trees ended up is anyone's guess but its thought that given the condition of the remaining plants, none or only a few would have had seed pods on them. as none were found on the plants controlled on this or the previous run.              



Weed update

September 2023

Very little weed control was undertaken this wet season, the flooding of the Fitzroy River ruled out the planned mesquite program planned for mid January.  However some mapping was undertaken from Camballin downstream to try and record the maximum flood height as it moved down the river. 

This was done on two separate days and the map below was created.  It was difficult to get a good handle on exact levels and their timing as gauging stations were been ripped out as the water moved downstream.  Potentially this map can be compared with LiDAR mapping data as it becomes available.

What this will give us is a high point of where weed seed and material could have been swept to as the flood moved through and will help with directing further survey work over the next few years. 

For the record the river in this survey at its narrowest point was six kilometres wide whilst its widest point down close to the mouth it was just under 30 kilometres wide.  The total area under water at that point can be seen on the map calculator below as 1,546 square kilometres.

In the East Kimberley however the prickly acacia program had its first control run for the year with an additional 6,000ha surveyed and only 13 plants found.  More importantly none were found outside the known infestation area.          


Rubber vine Update

December 2022

Two programs in the Kimberley are supported by the KRBA, one in the East and the other in the West Kimberley. In the East Kimberley program a total of 718 plants were treated off 246ha of land this year, down significantly from of 2,949 recorded in 2021 and the 13,684 recorded in 2020. An aerial survey was undertaken in March covering 33,424 hectares and located 27 seeders with nil plants found outside the known search area.

The West Kimberley program conducted its annual survey over 18,470 hectares with 151 flowering plants located in March. These and an unknown number of associated plants were controlled in May/June 2022. However during the mesquite survey undertaken in July (see above) a number of significant infestations were found along the Fitzroy River from 5 to 30 kilometres below the parameters of the currently known site. The full extent of these new infestations is currently unknown and will be determined in early 2023 by an extensive survey extending down the river as far as Camballin and perhaps beyond at this point.

These programs received funds for the work undertaken in 2022 and have received funds in the past as part of the Western Australian’s State Natural Resource Management Program.

Prickly acacia update

November 2022

Prickly acacia was first found in WA in 2002 with a lone plant spotted along the Duncan Road in the East Kimberley and in the following year a single plant was found in the quarantine yards just outside Kununurra. In addition to the quarantine yards find in 2003 an infestation southwest of Wyndham was found by a DPIRD Biosecurity Officer carrying out donkey control. This outbreak was spread across over 10,000 hectares with densities varying from scattered plants to very dense stands and was the largest infestation found outside Queensland at that time.

Whilst stock can often be the major cause for the distribution of seed in this case flood events seem to have been the biggest influencer of movement.

The KRBA provided some funding to the eradication program very early in its inception and took over the role of managing the infestation in 2017. Detailed data has been kept since 2010, in that wet season 9,230 plants were controlled from 5,050 hectares. In the 2012 wet season 14,680 hectares were surveyed and controlled but by that time overall plant numbers had reduced down to 582 and has been reducing incrementally since then. In the 2021 wet 227 plants were found out of 7,847 hectares searched.

The biggest issue with prickly acacia is the longevity of the seed produced. As an example on one site visited in April 2022 four immature plants were located where the last seeder plant was found in November 2010. Previous to 2022 a small plant was found on that location in 2016, another in 2014 and two in 2013.

The application of Graslan is largely responsible for the low numbers of seedlings found subsequent to the control of the original seeder plants but it’s clear that the dormancy of the seed long outlives the residual nature of the chemical in its initial application. The longevity of the seed and its ability to enable a small number of plants to germinate each year as a survival strategy for the species is apparent in the graph above with 83% of seedlings found to date removed in the first couple of years of the program and the remainder spread relatively evenly over the next ten years.

Currently it’s believed that seed viability can be up to 15 years for the plant. This means that its likely a couple of hundred plants will be found each year for the next four to five years over the entire site.

Fortunately after the initial knockdown of plants early in the program very few plants seeded due to the frequency of control work done and the practice of all seed pods removed from trees and surrounding soil and destroyed. The last verified seeding plant was found in 2018.

With the above in mind it is expected that old seeder locations, particularly isolated ones will have Graslan re-spread as a means to ensure germinating plants will die as soon as their roots find the residual chemical.

Diet of dingoes in the West Kimberley

November 2022

Linear clearings are frequently used by dingos to increase their mobility and landscape access. They will however sometimes modify the animal’s predator–prey relationships.

Tenaya Duncan, Patricia Fleming and Dr Stuart Dawson from Murdoch University, Western Australia recently carried out a year-long study on the diet of dingoes in the West Kimberley to test the hypotheses that the clearing of seismic lines would change the diet of dingoes in that area.

A total of 199 scats were collected from inside and outside the footprint of a seismic survey at three collection time points (once before and twice after the disturbance).  Overall, the diet of dingoes varied over seasons between control and treatment sites however there was no evidence of a shift in diet caused by clearing.

From the scat samples,

  • cattle were detected in 65.3% of samples,
  • agile wallaby were detected in 24.6%,
  • antilopine wallaroo in 5.5%
  • and most scats (67%) contained only one prey type.

Compared with the national average diet of cattle being present in 13% of samples for the arid and tropical regions of Australia, the consumption of cattle (live cattle and carrion) in the West Kimberley is very high, with greater consumption in the control and treatment sites at the end of the dry season compared to the early dry season the researchers noted. The concern is that if dingo numbers are not managed adequately there is huge potential for increased calf predation at the end of the dry season.

Feral cat and feral pigs were only present in one scat sample each, along with birds and reptiles in low frequencies.

For more information you can find the full article at www.publish.

Neem control trial summary

February 2022

A trial utilising encapsulated herbicides to control neem trees undertaken over a 12 month period has just been completed. Its scope was to –

  • Assess the effectiveness of encapsulated chemicals, including mortality rates and cost per unit of each treatment.
  • Assess the efficiency of use of the applicator tool.
  • Determine if control is possible at a reduced capsule rate.

The encapsulated chemicals used at full rate proved they could kill trees, though effectiveness in some instances ranged between 80 and 98%. The applicator used proved to be bulky and difficult to use in thickets of trees (note later versions are less bulky).

The trial identified that some of the encapsulated chemical rates could be halved and still be effective. Both Di Bak M and P provided similar effectiveness at a 50% application rate. Although the trial identified basal bark spraying as the quickest and most effective control option, encapsulated chemical injections do have a place in the weed control “toolbox”. For isolated weed control requiring a hike in on foot or transported by helicopter, it is easier to carry, and safer to transport encapsulated chemicals than sprayers for basal bark spraying.

A full summary of the trial can be obtained by emailing

State NRM funded weed programs

February 2022

The KRBA is a partner organisation in three weed programs trying to eradicate rubber vine and gamba grass infestations in a number of locations across the Kimberley. In 2021 these programs produced some mixed results but overall quite satisfactory progress has been made. Covid restrictions meant the usual contractor from the NT couldn’t be used on a couple of these programs and a coordinating contractor for one became unavailable at the last minute.

The East Kimberley rubber vine program ended up controlling 3,191 plants over 352 hectares with 137 considered to be breeders. This was down from the previous year’s total of nearly 14,000 plants from 685 hectares indicating the infestation is rapidly shrinking in terms of area covered and plants found.

The West Kimberley rubber vine program hit a speed bump with over 15,000 plants controlled of which 152 were breeders. This total was up considerably from the previous year of 961 plants. Most likely the increase came about from the 2020 survey having to be abandoned due to a lack of flowers because of the dry conditions at the time. In this survey program cameras detect flowering plants and the information is downloaded allowing the control team to find those plants.

In the East Kimberley gamba grass program only 13 plants were found, down from 23 plants found the previous year and 277 the year prior. It’s expected that within a couple more years the total should get down to zero plants found.

Ongoing funding for all these looks to be reasonably secure in the short term with both rubber vine programs gaining State NRM funding for 2022. The gamba grass program should have enough funds available for a couple of years and they may well see the control component through to completion. All programs receive and have received funds in the past as part of the Western Australian State Natural Resource Management Program.

Large feral herbivore program summary

July 2021

2020 was a reasonably successful year for our large feral herbivore cull program despite the initial delay due to COVID.  Numbers were down by 30% on the previous year due to the reduction in camel numbers reported. 

In total the numbers were 1,292 animals culled made up of donkeys, camels, some horses and pigs.  A total of 292 horses were removed from the Great Northern Highway between Kununurra and Halls Creek highway.  Previous to the work being undertaken there were a number of road accidents reported between horses and vehicles, particularly trucks along that stretch of road.  Ten new collars were deployed on donkeys including another three GPS collars that are assisting in determining the extent of the animal’s home range.  Additionally more genetic samples were collected as part of a project to see if there is any connectivity between separate donkey populations within the region. 

Rubber vine survey in West Kimberley breaks new ground

July 2021

The survey component of the rubber vine control program in the West Kimberly has been trialling cameras as a means to find adult flowering plants so they can be controlled before they seed.  During the proving process human observers were also used as a back up to the cameras.  This meant that the chopper had to be flown at speeds where those observers could see the plants without missing them.   In the recent survey the speed of the chopper was increased to up to 150kph.  At no more than 20 metres above ground level spotting flowers at this speed can be extremely hard.  However the cameras did an excellent job and easily met the standard hoped for with just over 80 plants found. Once the images are processed they are run through a software program designed to pick up the colour of the flower.  Those images are then passed on to volunteers who examine the image to determine if it is a flower or some other trigger such as a leaf reflecting light.  It should be noted that this project has received funding as part of the Western Australian Government’s State Natural Resource Management Program.

Weed control activities

July 2020

The annual prickly acacia wet season program in the East Kimberley was completed in February just prior to a week of good rains that will help with the uptake of the applied chemical.  Of the two control runs done over the wet a total of 95 plants were removed including just one seeding plant.  This is the second consecutive year the numbers of plants found has been under 100 and only a fraction of the 9,000 plus initially removed in 2010. 

All the seeds off the one plant were collected and destroyed on site. The East Kimberley rubber vine program started in early May with an initial aerial survey that covered approximately 30,0000 hectares, very few seeders were found in the existing control area however a new infestation covering approximately 13 hectares was found 11 kilometres away from another existing outlier infestation and 26 kilometres from the core infestation.  Estimates put the initial control at this location to take up to three weeks with follow up work for at least three to five years.

The West Kimberley rubber vine control program will start in July, the survey work due for earlier this year was suspended due to a lack of flowering.  The survey uses camera technology to identify rubber vine flowers by their distinctive colour so without flowers the survey cannot be run successfully.  In a normal year this method of detection has been demonstrated to have about a 30% higher success rate of finding plants compared to human observers.  

Vale Mick Everett

July 2020

Earlier this year long-time Kimberley shooter for DPIRD ‘Mick Everett’ sadly passed away in Perth.  Mick was a very popular figure in the region, particularly within the pastoral industry having been involved in aerial shooting programs since the early 1980’s.  He was the DPIRD coordinator for our ‘Judas’ donkey program in the Kimberley and at times would assist in a similar program in the Pilbara.   

A few years back he also got involved in rubber vine survey work in the West Kimberley with John Symanski’s Aquila program.  The annual survey uses cameras mounted on a chopper to find plants along the Fitzroy River but in the development of the techniques both human observers and the cameras were used in unison to properly test and refine those techniques. 

Mick’s last job for us was a five day cull in December last year of wild horses and camels south of Halls Creek under trying conditions.  

His knowledge of the local landscape and the people who operated within the region was huge and he was widely acknowledged as one of the good guys.  He will be greatly missed at both a professional and personal level right across the Kimberley. 

Kimberley Pig Research Project 2016 -2019

July 2020

In 2017 the KRBA provided funding through the Royalty for Regions program to the Australian Wildlife Conservatory (AWC) to expand an existing study looking at automatic feeder traps as a method of control for feral pigs.  That project came to an end in December 2019.  Following is a summary of the program and some of its results.    

There were five components to the research that were investigated over the duration of the project: 
  1. Investigate the best method for attracting feral pigs to bait stations and traps. 
  2. Develop and test an automated bait station that can be transported by helicopter, used in remote areas and left for long periods unattended. 
  3. Develop and test trap set-ups suitable for helicopter slinging for use in remote areas. 
  4. Gather data of pig damage throughout the year at specific swamps, before and after pig control efforts. 
  5. Gather and analyse location and habitat data using GPS collars, to determine -  
  • How far pigs travel and how this varies with seasonality. 
  • Home range size and overlap for male and female pigs. 
  • Habitat requirements and how behaviour changes throughout the day and night. 
  • What are the drivers for pig presence in the Kimberley.   

Automated feeders were constructed and erected in areas, across a range of pig densities.  All feeders had two cameras facing the feeder to identify if a subset of pigs would come in to the feeder but not feed from it.  In addition, a feeding experiment was conducted at a station’s bone pit, where cameras were deployed on both the automated feeder and pit to examine which food source was preferred.     A total of 20 feral pigs were fitted with GPS collars in three areas to encompass a latitudinal spread across the rainfall gradient and to enable research on differing land uses (pastoral and properties managed for conservation).  All collars ran for more than 2 years after deployment. 

 A trial was undertaken using the placebo version of the commercially produced “Pig-out” bait.  While the baits seemed palatable their use was ruled out as they are prone to drying out when exposed to the air and are relatively expensive.  A previous study in the Kimberley trialled the use of 1080 laced grain, which was found to be very successful.  At the time of the project lacing grain with 1080 to target pigs was not permitted however this was changed in 2019 and is now permitted under the code of practice for the safe use of 1080, PAPP & strychnine. This means approved pastoralists can now access 1080 concentrate for the purposes of lacing grain to target pigs.  

 Two experiments were undertaken to look at different ways of introducing poison to pigs.   First was a trial using grain, offered in piles adjacent to the feeder to see how much grain would have been consumed by pigs.    This would help calculate how much laced grain will be required when undertaking trials using poison.  The largest group of pigs observed in a single event did not exceed 15 animals.  It is estimated that to kill a group of this size at least 10 kg of laced grain should be deployed.  Also trialled was the use of gel capsules dispersed through heaped grain. The idea being that the capsules were small enough to be consumed along with the grain.  Initial results were promising as pigs consumed the capsules but it’s unlikely that this method could be used in the short term as it will need more evidence to be potentially registered as an approved method.   

 In brief, the research has achieved the following –

  • Successful development of an automated baiting system that can be deployed for control of feral pigs in remote parts of the Kimberley, and optimised the delivery of baits using that system. 
  • Identification of movement patterns of pigs in the central and western Kimberley, including differences in habitat use across the rainfall gradient from north to south.  

Modelling conducted by AWC shows that suitable habitat for feral pigs is much larger than the area currently occupied and pigs are still expanding throughout the Fitzroy Basin and in the North Kimberley, there is a significant potential for the spread of feral pigs due to the high rainfall and the many small rivers that dissect the region.  

 Feral pig management in the north will be complex as pigs are likely to be dispersed throughout the area due to the abundant water.  In the south, however, pigs are largely constrained to the larger watercourses, particularly in the drier parts of the year. 

Weed work summary

December 2019

Prickly acacia seedsWork continues on the prickly acacia infestation that is scattered over 10,000 hectares to the west of Wyndham.  Less than 100 plants were found and controlled on the site in October of this year.   Recent fires made the work difficult with many plants burnt or just half burnt meaning some plants may not take up the chemical applied.  Only two seeding plants were found this year, the seed pods from those plants were removed and deep buried on site to ensure that the wild cattle in the area would not eat the seed and cause its spread.  

Rubber vine plant numbers controlled in both the East and West Kimberley continues to decline.  In the east 3,471 plants were located, just over half the plants found in the previous year.  In the west just 728 plants were found of which only 16 vines were mature.  

The KRBA’s Field Officer “Blu’ carried out a survey of grader grass in the North Kimberley.  In many locations the density and extent of the plant had increased significantly since the North Kimberley LCDC survey in 2011.  In some spots the movement along roads was measured at more than 20 kilometres in the eight years between surveys.   The KRBA will be carrying out field trials in conjunction with some land managers across the Kimberley this wet season to gauge the effectiveness of new chemicals and methods of application on grader grass.  Additionally we will be keeping an eye on similar work being done in the NT looking at other means of control.