Achieving no net loss in habitat offset of a threatened frog required high offset ratio and intensive monitoring


of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
The use of habitat offset to mitigate the impact of development on threatened species is becoming increasingly popular. Despite a plethora of theoretical work on the requirements of habitat offset to achieve no net loss, there are very few examples
  Achieving no net loss in habitat offset of a threatened frog required high offsetratio and intensive monitoring Evan J. Pickett ⇑ , Michelle P. Stockwell, Deborah S. Bower, James I. Garnham, Carla J. Pollard, John Clulow, Michael J. Mahony School of Environmental and Life Sciences, University of Newcastle, University Drive, Callaghan, NSW, Australia a r t i c l e i n f o  Article history: Received 23 July 2012Received in revised form 7 September 2012Accepted 21 September 2012 Keywords: Habitat offsetMitigationRestorationUncertaintyAmphibianNo net lossMultiplier a b s t r a c t The use of habitat offset to mitigate the impact of development on threatened species is becomingincreasingly popular. Despite a plethora of theoretical work on the requirements of habitat offset toachieve no net loss, there are very few examples of successful habitat offset programs and monitoringregimes to detect success. We present a case study of a population of the threatened green and goldenbell frog ( Litoria aurea ) which was impacted by urban development through the removal of nine ponds.Development was concurrent with habitat offset and construction of a large number of ponds whichresulted in a 19-fold increase in available pond area. Through the use of mark recapture surveys, the pop-ulation size was determined pre- and post-development. Despite the creation of ponds in the immediatevicinity of the development there was a decrease in the pond area and a measured decline in the popu-lation located within the area where the development occurred. However, the overall pond constructionprogram also involved the addition of considerable habitat away from the immediate vicinity of thedevelopment which resulted in a 19-fold increase in pond area and an approximate 1.2–3.5-fold increasein population size. No net loss in population size to 95% confidence was achieved only when including allpond construction. This study demonstrated that to achieve no net loss for a habitat offset program canrequire extensive levels of habitat creation with intensive monitoring to detect it. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction Loss and alteration of habitat has seen the reduction of species atthe local, nationaland global scaleand these factorsare listed asthemostcommoncauseofspeciesdecline(Butchartetal.,2010).Thoughthelarge-scaleclearingofnaturalhabitatforagriculturehasrecentlydeclinedinmanydevelopedcountries,industrialandurbandevelop-ment continues to endanger many species and habitats over a widegeographical area (McKinney, 2002; Pauchard et al., 2006).Habitat loss mitigation through the creation of new habitat hasbeenanincreasinglypopularrequirementfordevelopmentapprov-als (Edgar et al., 2005; Madsen et al., 2010); wetland restoration orcreation in the US alone increased from 7148 ha to 56,613 ha from1992to2002(tenKateetal.,2004).Theintentionofhabitatoffsetisto achieve ‘no net loss’ or ideally lead to a ‘net gain’ in the conser-vation value of an area impacted by development (Quintero andMathur, 2011). For habitat offset concerning a single threatenedspecies, this usually means no loss in population size or viabilitythrough the actions of a development. Successful implementationof habitat offset enables infrastructure projects to contribute toconservation efforts through mitigation programs, whilst long-term monitoring programs to evaluate success can provide muchneeded insight into the population dynamics of threatened speciesand communities (Quintero and Mathur, 2011).The effectiveness of habitat offset has been widely debated, asthe quality and extent of offset and level of monitoring and revieware often insufficient to ensure that successful offset has beenachieved (Maron et al., 2012; Matthews and Endress, 2008; Morriset al., 2006). The creation of habitat is made difficult by the level of uncertainty in the eventual outcome of the program. Thoughcreated habitat can resemble the composition of existing habitat,certain ecological processes can be difficult to restore, possiblyreducing the compatibility for the target species or community(Moreno-Mateos et al., 2012). A time lag is also expected betweenthe creation of habitat and habitation by the target species, assome habitat resources require later-stage succession (Moilanenet al., 2009; Vesk et al., 2008; Zedler, 1996). This can result in somedevelopments proceeding before the offset habitat has the capacityto achieve no net loss. This time lag is pronounced in certainhabitat such as woodlands and some grassland, but can be rapidin highly dynamic or transient systems, such as mudflats, saltmarshes and freshwater wetlands (Morris et al., 2006). 0006-3207/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. ⇑ Corresponding author. Tel.: +61 249212045; fax: +61 249215472. E-mail address: Pickett).Biological Conservation 157 (2013) 156–162 Contents lists available atSciVerse ScienceDirect Biological Conservation journal  The uncertainty of success for the development of offset habitathas resulted in some broad recommendations for its implementa-tion.Twoof the major recommendations concernthe size andloca-tion of habitat offset projects as a means of increasing theprobability of creating the ecological processes required for suc-cess. A high offset ratio, where more habitat is created than lost,is recommended for species with a risk of failure (Bruggeman etal., 2005; Dunford et al., 2004; Moilanen et al., 2009). Under thiscircumstance, a small proportion of success within created habitatmay still achieve no net loss as a large quantity of habitat iscreated. The second recommendation is to build offset at a closeproximity to the lost habitat in an attempt to maintain the srcinalcomposition, increase the probability of colonisation and to incor-porate localised habitat characteristics or ecological processes(Moilanen et al., 2009). The final recommendation is to delaydevelopment so as to allow succession of offset habitat to achieveno net loss. However, the slow succession of some environmentsand the economic value of some developments to society meanthat many developments proceed before this is achieved, andtherefore management of the offset habitat is required to ensuresuccessful mitigation (Morris et al., 2006).The literature contains extensive theoretical justifications per-taining to the above recommendations (seeMorris et al. (2006)for a summary). However, criticisms of habitat offset programs in-clude that there is a consistent failure to monitor and report thesuccess of offset (Edgar et al., 2005), and that success is frequentlyevaluated based on excessively lenient criteria (Matthews andEndress, 2008). Monitoring of habitat offset projects is requiredpre- and post-development to determine success, and long-termmonitoring is required to evaluate sustainability of the population(Quintero and Mathur, 2011). A review of great crested newt( Triturus cristatus ) habitat offset projects in the UK found that just49% of projects included a post-development monitoring period.Furthermore, the average length of this monitoring period lasted1.8 years, which would not account for any negative effects thatsuccession may have on the population (Edgar et al., 2005).We present a case study of a threatened species for habitat off-set that was successful in achieving no net loss through the crea-tion of large areas of habitat. This could be successfully evaluatedwith the use of long-term data that was collected for the targetpopulations prior to and after a development that resulted in theloss of habitat. This case study highlights the complexity of dealingwith habitat offsets for a species which is perceived to be ‘straight-forward’ based on its biology and habitat requirements (see Sec-tion2.1), and demonstrates that the level of effort required tosuccessfully construct and monitor habitat offset may be drasti-cally underestimated for most infrastructure projects. 2. Materials and methods  2.1. Study species and site The green and golden bell frog ( Litoria aurea ) is native to thesouth-eastcoast of Australia and is listed as vulnerableby the IUCN(Hero et al., 2004). Populations have declined since the 1970s, con-tracting towards the coast with just 37 populations occurring inthe state of New South Wales. This coastal contraction has placedthe remaining populations of  L. aurea under increased threat fromurban development (White and Pyke, 2008b). L. aurea has been ob-served to rapidly inhabit ponds after creation and has the highestrecorded fecundity for a native Australian frog (Hamer and Mah-ony, 2007). These traits make L. aurea a perceived ideal candidatefor habitat offset as habitat can be rapidly created and inhabited.One of the largest populations of  L. aurea is found at SydneyOlympic Park, the site of Australia’s biggest urban remediation pro- jects (Darcovich and O’Meara, 2008). L. aurea was historicallyfoundthroughout the park, including within a disused quarry, known asthe Brickpit, which was conserved to maintain its population of  L.aurea. Long-term monitoring has been commissioned by the Syd-ney Olympic Park Authority throughout the development periodand has been maintained through the post-development period.A development occurred in the Brickpit in 2000 which resultedin the loss of 9 of 26 ponds by flooding two lower levels of thequarry to create a water reservoir (Australian Museum BusinessServices, 1999). This equated to a loss of 3351 m 2 of pond surfacearea and 775 m of pond edge. As a mitigation measure, 19 pondswere constructed within the Brickpit. An additional 24 ponds wereconstructedthroughoutSydney Olympic Park as part of the L. aurea management plan to conserve the population outside the Brickpit(Fig. 1). A requirement for any development in the Brickpit wasthat these external ponds were successfully colonised by L. aurea (Darcovich and O’Meara, 2008). These changes equated to the cre-ation of 2249 m 2 of pond area in the Brickpitand 64,757 m 2 in totalthroughout Sydney Olympic Park (830 m and 6927 m of pond edgerespectively;Table 1). These ponds were created within 2 km of the Brickpit, on top of historical locations for the species to removethe issue of proximity of offset habitat to removed habitat. Offsethabitat outside the Brickpit was also created adjacent to alreadyoccupied ponds.This study focused on two major offset areas outside of theBrickpit where L. aurea exhibit the highest abundance known asthe Northern Water Feature and Narawang Wetland. It also in-cludes a subset of the Brickpit ponds where abundance was high-est, including most offset ponds within the Brickpit.  2.2. Monitoring  Monitoring of the population was conducted by differentgroups during the life of the project, resulting in variable methodsand level of effort. These methods included auditory surveys, tad-pole surveys, timed visual encounter surveys to determine relativeabundance and mark recapture surveys. We have analysed themark recapture data so as to determine the population size of the Brickpit and offset habitat wherever this data was available.Mark recapture involved repeated surveys of ponds where frogswere captured with a disposable plastic bag to prevent diseasetransmission. Frogs were scanned to detect a passive integratedtransponder (PIT) tag, and newly encountered individuals weremarked via subcutaneous insertion of a PIT tag in the dorsolateralregion of the body and were then released at the site of capture.Regular closed-population mark recapture surveys were con-ducted annually in the Brickpit from 2007 to 2011. Developmentof the Brickpit occurred from August 1999 to June 2000. Twoclosed-population mark recapture surveys were completed 9 and6 months prior to the beginning of development within the brick-pit. During the initial stages of development, frogs were removedfrom the development area to limit direct mortality of frogs. Thesefrogs were relocated to ponds adjacent to the development area,and a single mark-recapture survey was conducted concurrentlywith this removal process. A single mark recapture survey was alsoconducted 10 months after completion of the development.All surveys within the brickpit were conducted to follow theassumptions of the closed population model (Pollock et al.,1990). Consistent closed-population mark recapture surveys con-form to the Pollock’s robust design model which incorporates sam-pling at two temporal scales, known as primary and secondarysampling events (Kendall, 2001; Pollock, 1982). Primary samplingevents were separated by long intervals at which migration, deathand recruitment occur (open population). Within each primarysampling event, more than one secondary sampling occasion oc-curred over a short period during which the population can be as- E.J. Pickett et al./Biological Conservation 157 (2013) 156–162 157  sumed as closed. By incorporating closed population estimates forabundance and open estimators for survival within the one model,the overall analysis is more robust than if these were estimatedseparately (Kendall, 2001). Pollock’s robust design estimates:  Apparent survival ( u ) – probability an animal stays within thestudy area between primary surveys. Removal of an animalcan occur through mortality or permanent emigration.  Temporary emigration ( c ) – probability an animal migratesaway from the survey area for at least one sampling occasionand subsequently migrates back.  Capture probability (  p ) – probability an individual is capturedduring a survey period  Recapture probability ( c  ) – probability a marked animal is cap-tured during a survey period.  Population size ( N  ) – size of the target population.The assumptions of Pollock’s robust design are as follows:  Capture and survival probability of each individual is indepen-dent of other individuals.  No births, deaths or migration occur during secondary samplingoccasions (closure).  Survival probabilities are equal for all individuals in thepopulation. Fig. 1. Map of ponds at Sydney Olympic Park representing the ponds lost during the Brickpit development (black), existent ponds unaffected by development (dark grey) andponds created for habitat offset program (light grey).  Table 1 Habitat offset ratios for the Brickpit development site population and the totalpopulation across Sydney Olympic Park using different descriptions of habitat:num-ber of distinct ponds, surface area of ponds and length of pond edge. Pond number Surface area Pond edgeBrickpit 1:2.1 1:0.7 1:1.1Total park 1:4.7 1:19.3 1:8.9158 E.J. Pickett et al./Biological Conservation 157 (2013) 156–162   Marks are unique and not lost or misread.  Capture and marking do not affect survival or recapture rate.  Marked individuals are a true representation of the population.Robust design mark recapture surveys were also conducted forthe Northern Water Feature and its surrounding ponds.Competing models were tested by comparing the effect of keep-ing each parameter constant with varying parameters over time,and by determining whether capture and recapture probabilitieswere equal during a single sampling period. Where appropriate,capture and recapture probability was also modelled against thenumber of people in each survey, the length of the survey (re-corded as number of nights) and the maximum temperature inthe day preceding the survey from the Sydney Olympic Park Bu-reau of Meteorology weather station.Model selection was based on Akaike’s Information Criterionwith correction for small samples (AICc) produced by programMARK. The most parsimonious model was determined as the mod-el with the smallest AICc value. The D AIC value is the differencebetween a model and the most parsimonious model, and D AIC val-ues of less than two could not be considered different enough toreject (Burnham and Anderson, 2002). To remedy this, each modelwas weighted according to the D AIC value and the parameter out-puts were averaged according to the methods of Burnham andAnderson (2002). Models that failed to converge were removedfrom the candidate model set to prevent influence on the modelaveraging results.Surveys to determine the relative abundance of  L. aurea at eachpond were undertaken consistently since 1996, and marking of individuals was introduced to these surveys in 2008. This markingdata from 2009 to 2011 was used to determine population size of the third high-abundance offset habitat area within Sydney Olym-pic Park, the Narawang Wetland, using robust design mark recap-ture analysis. However, the assumption of closure was unlikely tobe met for the secondary sampling periods as they occurred1 month apart which was likely causing slight positive bias forpopulation size in the Narawang Wetland (Kendall, 1999).Population size was not estimated for existing habitat outsidethe Brickpit, mostly due to the low density of  L. aurea within theseponds. Therefore, evaluation of no-net-loss is based on theassumption that the development within the Brickpit and the cre-ation of offset habitat did not have a negative impact on the pop-ulation of  L. aurea within existing ponds outside the Brickpit.For the purpose of this study we only used population size as itis a measure of ‘no net loss.’ In total, five mark recapture modelswere produced: three for the brickpit (1999–2000, 2000–2001and 2007–2011) and one each for the Northern Water Featureand Narawang Wetland (2009–2011).To determine whether no net loss was achieved, the populationsize of the Brickpit based on the 1999–2000 pre-development sur-veys was compared to the combined population size of the Brick-pit, Northern Water Feature and Narawang Wetland in 2010. Theupper 95% confidence interval of the population pre-developmentwas used as the threshold of success whereby the lower 95% con-fidence interval of the post-development population size estimatehad to reach this level so as to remove the issue of uncertainty. 3. Results Effort in the mark recapture surveys within the Brickpit devel-opment site increased over the 13 years of study (Table 2). There-fore, the most parsimonious models for the five surveys weredifferent (Table 3). The Brickpit mark recapture indicated a popu-lation decline from 276–551 in 1998 to 131–156 in 2010 (95% con-fidence intervals;Fig. 2), representing an approximate 1.5–2.8-foldloss in population size. However, the level of temporal variabilityin population size is high within the Brickpit, as is demonstratedby an approximate 100% increase between 2010 and 2011. The ob-served decline may therefore be an artefact of limited sampling asthe range of population sizes were unknown prior to developmentwithin the Brickpit.Areas of highest abundance within the offset habitat supportedlarge populations according to the mark recapture analysis. TheNorthern Water Feature and the Narawang Wetland were foundto have populations that ranged between 182–208 and 252–492(95% confidence intervals) respectively.Comparison of population size of the Brickpit prior to develop-ment and the total population size after development indicatesthat the offset program has been successful in achieving no net losswith an approximate 1.2–3.5-fold increase in population size. Thelower confidence interval post-development exceeds the upperconfidence interval pre-development and the threshold for successwas therefore met (Fig. 3). 4. Discussion The habitat offset program at Sydney Olympic Park was exten-sive in its attempt to achieve no net loss, with a 19-fold increase inpond area and 8.9-fold increase in pond edge. Despite these largeincreases in available habitat, there was not an equivalent increasein population size with an approximately 1.2–3.5 fold increase inpopulation size. Despite some offset attempts within the Brickpitdevelopment site, an overall loss of habitat resulted in a declinein population size within the Brickpit.The disparity between the amount of created habitat at SydneyOlympic Park and the increase in population size could indicatethree different processes: L. aurea (1) was still in the process of col-onising available habitat in 2010, (2) inhabited the offset habitat ata lower density or (3) could not colonise all available areas. L. aurea has the highest recorded fecundity of any Australian frog and hasbeen noted for its ability to rapidly colonise ponds after their cre-ation (Goldingay and Newell, 2005; Hamer and Mahony, 2007). Itis therefore unlikely that the colonisation process extended the en-tire decade of this study making the first explanation unlikely.Created wetlands have been shown to experience lower vegetationstructure, biodiversity, invertebrate assemblages and productivitythan natural wetlands (Brown et al., 1997; Moreno-Mateos et al.,2012; Stanczak and Keiper, 2004). These factors could result in alower density for L. aurea or if productivity is too low could resultin unviable populations over the long-term. Design of ponds forhabitat offset was based on the knownhabitat featuresfor this spe-cies (Darcovich and O’Meara, 2008; Pyke and White, 1996),although it is possible that an obscure habitat feature has not been  Table 2 Summary of captures for each mark recapture survey. Captures are the number of frogs captured during a survey, with the number of recaptures indicating how many of theseanimals were previously captured in the survey. 1998 1999 2000 2001 2007 2008 2009 2010 2011Captures 156 96 89 69 79 122 193 206 301Individuals 135 87 79 61 69 103 135 142 221Recaptures 21 9 11 8 10 19 71 88 103 E.J. Pickett et al./Biological Conservation 157 (2013) 156–162 159  identified as relocations for this species are often unsuccessful(Stockwell et al., 2008; White and Pyke, 2008a). This would resultin uninhabited areas where this unknown habitat feature was ab-sent. This last point is possible due to the unknown role of habitatin controlling pathogens; particularly the chytrid fungus which isknown to affect this species and population (Penman et al., 2008).Determining the threshold for success is a shortfall for manyoffset programs (Matthews and Endress, 2008), but is simplifiedfor single species projects which can use estimates of populationsize. The threshold for success was reached for this offset projectwhen offset outside of the Brickpit was included, and this thresh-old incorporated uncertainty in the population size estimate. Thepre-development population size estimate ranged from 276 to551 (95% confidence intervals). In order to achieve no net loss, itwasalso necessary for the population to have increased ratherthanremaining the same after development occurred, as the largeamount of uncertainty increased the upper confidence limit whichwas used as the threshold for success. Ideally, a large amount of ef-fort should be invested in the initial phase of habitat offset projectswhen the pre-development population size is assessed, so that thethreshold for success can be determined more precisely. Monitor-ing regimes should be explicit in the minimum level of uncertaintyand flexible so as to allow increased monitoring until this level isreached.Determining the success threshold is complicated by temporalvariability in population size. Determining population variabilityrequires long-term data, as rare impacts on population size suchas 1-in-50 year floods, cannot be detected without repeated sur-veys (Pimm and Redfearn, 1988). It is common practice for thedevelopment process to be rushed, but attempts should be madeto incorporate as many pre-development population size estimatesas possible so as to determine some level of population variability.Temporal variability was evident with this population, as the markrecapture results indicated a highly variable population size withan approximately 100% increase between 2010 and 2011. Failureto incorporate long-term data could result in a population size esti-mate that is extreme for that population and not a representationof the norm.Use of a single metric for success is likely to be an over-simpli-fication of the viability of a population subject to habitat loss andoffset (Traill et al., 2007). The habitat offset program at SydneyOlympic Park resulted in an increased population size over a largerdistribution. This could have both positive and negative impactsfor population viability, particularly for populations acting as mul-tiple metapopulations. Creation of offset habitat that is furtherfrom the existing population than the lost habitat would likely re-duce the level of migration. If migration is substantially reduced,the capacity for nearby populations to be recolonised after extinc-tion is diminished. Alternately, an increased spread of the popula-tion will mean localised events that negatively impact ametapopulation will have less impact on the population as a whole(Hanski, 1999). The impact on habitat offset increasing the distri-  Table 3 The most parsimonious models from each candidate model set for mark recapturesurvey. Comparison of models for parsimony used Akaike’s Information Criterion(AICc) and models with D AICc < 2 are included within the table. Model parametersinclude: apparent survival ( u ), temporary emigration ( c 00 and c 0 ), capture probability(  p ) and population size ( N  ) and included the effects of time ( t  ), the number nights asurvey lasted (surveyNights), the mean number of people in the survey over thenumber of survey nights (meanSurveyors). Some parameters did not vary (.) andsome were constrained as indicated by the ‘‘=’’ symbol. Model D AICcNarawang u ( t  ) p ( t  ) = c  c 00 = 0 c 0 = 0 N  (.) 0Northern Water Feature u (.) p ( t  ) = c  c 00 = 0 c 0 = 0 N  ( t  ) 0Brickpit 1999–2000 u (.) p (surveyNights) = c  c 00 = 0 c 0 = 0 N  ( t  ) 0 u (.) p (surveyNights) = c  c 00 = 0 c 0 = 0 N  (.) 1.7Brickpit 2000–2001 u (.) p ( t  ) = c  c 00 = 0 c 00 = 0 c 0 = 0 N  ( t  ) 0Brickpit 2007–2011 u ( t  ) p (meanSurveyors) = c  c 00 (.) c 0 (.) N  ( t  ) 0 u ( t  ) p ( t  ) = c  c 00 (.) c 0 (.) N  ( t  ) 1.76 Fig. 2. Size of the Brickpit population over time. Dotted line indicates beginning of development pressures between August 1999 and June 2000. Note that frogs weremoved from development site in 2000–2001. Error bars indicate 95% confidenceintervals. Fig. 3. Population size of the Brickpit development site before developmentcompared to the combined populations of the Brickpit and two offset areas(Northern Water Feature and Narawang Wetland) 10 years after development.Despite a population decline within the Brickpit, the overall habitat offset programproduced no net loss in the population size. Error bars indicate 95% confidenceintervals.160 E.J. Pickett et al./Biological Conservation 157 (2013) 156–162
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks