DOI10.1068b31134 Predicting ecological connectivity in urbanizing landscapes
Application of a Random Forest algorithm to predict spatial distribution of the potential yield
Ecological Modelling 222 (2011) 1471–1478Contents lists available at ScienceDirectEcologicalModellingj o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /e c o l m o d elApplication of a Random Forest algorithm to predict spatial distribution of the potential yield of Ruditapes philippinarum in the Venice lagoon,ItalySimone Vincenzi a ,∗,Matteo Zucchetta b ,Piero Franzoi b ,Michele Pellizzato c ,Fabio Pranovi b ,Giulio A.De Leo d ,Patrizia Torricelli baDipartimento di Scienze Ambientali,Universitàdegli Studi di Parma,Viale berti 33/A,I-43125Parma,ItalybDipartimento di Scienze Ambientali,Informatica e Statistica,UniversitàCa’Foscari Venezia,Castello 2737/B,30122Venezia,Italy cAGRI.TE.CO sc Ambiente Progetto Territorio,Via Carlo Mezzacapo 15,30175Marghera,Italy dDipartimento di Scienze Ambientali,Universitàdegli Studi di Parma,Viale berti 11/A,I-43125Parma,Italya r t i c l e i n f o Article history:Received 25June 2010Received in revised form 20January 2011Accepted 6February 2011Available online 4 March 2011Keywords:Ruditapes philippinarum Venice lagoon Random Forest YieldHabitat suitabilitya b s t r a c tWe present a modelling framework that combines machine learning techniques and Geographic Infor-mation Systems to support the management of an important aquaculture species,Manila clam (Ruditapes philippinarum).We use the Venice lagoon (Italy),the first site in Europe for the production of R.philip-pinarum ,to illustrate the potential of this modelling approach.To investigate the relationship between the yield of R.philippinarum and a set of environmental factors,we used a Random Forest (RF)algo-rithm.The RF model was tuned with a large data set (n =1698)and validated by an independent data set (n =841).Overall,the model provided good predictions of site-specific yields and the analysis of marginal effect of predictors showed substantial agreement among the modelled responses and available ecolog-ical knowledge for R.philippinarum .The most influent environmental factors for yield estimation were percentage of sand in the sediment,salinity,and water depth.Our results agree with findings from other North Adriatic lagoons.The application of the fitted RF model to continuous maps of all the environmen-tal variables allowed estimates of the potential yield for the whole basin.Such a spatial representation enabled site-specific estimates of yield in different farming areas within the lagoon.We present a pos-sible management application of our model by estimating the potential yield under the current farming distribution and comparing it to a proposed re-organization of the farming areas.Our analysis suggests a reduction of total yield is likely to result from the proposed re-organization.© 2011 Elsevier B.V. All rights reserved.1.IntroductionThe Manila clam Ruditapes philippinarum (Adam and Reeve,1850),which is of Indo-Pacific origin,was introduced in the Venice lagoon (Fig.1)in the 80s as a culture species (Cesari and Pellizzato,1985)and radically changed the exploitation of living resources in the lagoon.Within a few years,R.philippinarum became the most important exploited species in the lagoon,with a production reach-ing a peak of over 40,000t y −1at the end of the 90s,estimated from various sources and using expert knowledge by Pellizzato and Da Ros (2005).No official fishery landings data for the whole lagoon is available and yield potential is largely unknown,despite the relevant social,economic and environmental consequence of the exploitation activities.Since the introduction,the exploitation of R.philippinarum has been carried out in a regime of free access.In 1999,the Province∗Corresponding author.Tel.:+390521905696;fax:+390521906611.E-mail address:simone.vincenzi@nemo.unipr.it (S.Vincenzi).of Venice began a gradual shift to a concession regime,i.e.,to a system where harvesting areas are divided by the regulatory agency among a number of concessions,each managed by local clam fishermen under a strict set of rules on access limitation and exploitation effort.Technically,concessions are divided in farm-ing (i.e.,where clams are seeded)and fishing (i.e.,where clams are naturally recruited)areas.In the following,we will use the word concession without further differentiating between farming and fishing areas.In 2007,about 42km 2of the Venice lagoon were given in concession to fishermen for harvesting of R.philippinarum (Fig.2a and Table 1).However,the transition from uncontrolled fishing to a “culture-based fishery”based on correct and sustainable rearing procedures,while being successful in reducing production of R.philippinarum ,revealed to be more complex than expected and cannot be con-sidered successfully completed (Pellizzato and Da Ros,2005).The Province of Venice is willing to reduce the number of fisher-men operating in the lagoon (from about 900to 600)and to remodel and reduce the areas given in concession to clam fish-ermen (G.R.A.L.,2006,2009;Province of Venice,2009)to reduce0304-3800/$–see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.ecolmodel.2011.02.0071472S.Vincenzi et al./Ecological Modelling222 (2011) 1471–1478Fig.1.Map of the Venice lagoon (Italy).The basin can be divided by two watersheds in three main sub-basins,Northern,Central andSouthern.Fig.2.Areas given in concession by local authorities in 2007(Panel a)and,according to the reshaping plan proposed by the Province of Venice,in 2013(Panel b).Panel c shows no-takes zones for sanitary reasons (sites highly polluted)and sites of conservation concern (seagrass meadows).health risks linked to industrial pollutants or urban waste,mini-mize the environmental impacts of bottom dredging,such as the loss of sediments (e.g.,Molinaroli et al.,2007),increase of water turbidity and movements of nutrients and pollutants (Pranovi et al.,2004;Sfriso et al.,2005),protect habitats of conservation concern,such as seagrass meadows (Fig.2c),and maximize pro-duction in order to minimize fishing effort,both in space and time.S.Vincenzi et al./Ecological Modelling222 (2011) 1471–14781473Table1Yield predicted by the Random Forest model for the whole lagoon of Venice,thethree sub-basins and concession areas for2007and2013.Total yield for areas givenin concession is expected to decrease after the remodelling plan(2013).Total yield (t y−1)Area(km2)Yield per unit area(g m−2y−1)Whole basin69,883216.30323.08Northern sub-basin14,74383.60176.35Central sub-basin41,67999.00421.00Southern sub-basin13,46133.70399.44Harvesting areas200722,73142.59533.72Harvesting areas201318,27735.52514.54Harvesting areas2007North1,117 5.98186.71Harvesting areas2007Central17,70626.56666.65Harvesting areas2007Southern3,90810.05388.87Harvesting areas2013North3,3168.38395.75Harvesting areas2013Central11,75916.31720.97Harvesting areas2013Southern3,86210.45369.57In this context,the identification of suitable harvestable grounds and a reliable estimation of site-specific commercial yield poten-tials are necessary to guarantee a sustainablefishery,to improve economic efficiency of clam farming,ensure an equitable share of exploitable areas to competing subjects interested in the exploita-tion of R.philippinarum and to foster transparency in the decision making process aimed at planning the future exploitation activities.Habitat suitability(HS)models or models predicting species distribution(the two definitions will be used interchangeably) constitute good tools supporting decision-making within the framework of applied biology.HS models have been often used to improve our understanding of species–habitat relationship in space and time and to predict the likelihood of occurrence and abun-dance of a species using habitat attributes affecting its survival, growth and reproduction(e.g.,Guisan and Thuiller,2005;Hirzel et al.,2006;Santos et al.,2006).Habitat suitability approaches have also been used for identifying appropriate sites for mollusk farming in North-America and Mexico(e.g.,Kapetsky et al.,1988; Aguilar-Manjarrez and Ross,1995).Vincenzi et al.(2006a,b,2007) developed simple HS models for the estimation of yield potential of R.philippinarum in the Sacca di Goro lagoon(North Adriatic,Italy) by using semi-empirical and zero-inflated regression models.In recent years,machine learning methods,such as classification and regression trees(Dˇz eroski and Drumm,2003;Seoane et al.,2005) artificial neural networks(ANN,Pearson et al.,2002;Dedecker et al.,2004)and Random Forests(Benito Garzón et al.,2007,2008) have been proposed for the development of spatial distribution models.Machine learning methods are capable of detecting com-plex relationships among model variables without making a priori assumptions about the type of relationship,such as a linear depen-dence on predictors,and are able to process complex and noisy data (Recknagel,2001).In this work,we used a Random Forest algorithm(Breiman, 2001)to explore the relationship between the yield potential of R. philippinarum and several environmental factors deemed impor-tant for the occurrence and abundance of the species.Several studies have shown that Random Forest(RF)models,based on an automatic combination of tree predictors,often reach top predic-tive performances compared to other methodologies(e.g.,Prasad et al.,2006;Cutler et al.,2007).Our paper is organized as follows: after a brief description of the study area,of the environmental fac-tors linked to the occurrence and abundance of R.philippinarum and of available data,we briefly illustrate the main features of the Ran-dom Forest model and proceed with the calibration and validation of the model by using two independent data sets relative to year 2007.Then,we apply the Random Forest model to the Venice lagoon to obtain estimates of potential yield inside and outside the areas given in concession in2007.In addition,we predict the yield poten-tial of R.philipparum in areas that,according to the remodelling plan proposed by local authorities,will be given in concessions in2013. Finally,we discuss the relevant features,limitations and further development of the Random Forest approach.2.Materials and methodsThe resolution(i.e.,operational scale)chosen for the study was 100×100m cells(site),for a total of45,443cells.2.1.Study areaThe Venice Lagoon is located in the Northern Adriatic Sea and is the largest lagoon in the Mediterranean basin,with an area of about 550km2including emerged lands(Fig.1).The Venice lagoon is a shallow coastal ecosystem(average depth1.2m,Molinaroli et al., 2007),where large areas,covering about75%of the total surface (Molinaroli et al.,2009),are connected by a network of channels, whose depth is mostly<2m(Solidoro et al.,2002).Deeper channels connected to three wide mouths(Lido,Malamocco and Chioggia) maintain the lagoon–sea communication and allow tidalflows to enter the lagoon,within a range of±50cm during spring tides (Umgiesser et al.,2004).The basin can be divided by two water-sheds in three main sub-basins(Northern,Central and Southern, Fig.1)(Solidoro et al.,2004).The bottom sediments of the basin consist mainly in clayey silt,with a mean mud content of about 80%in dry weight,showing a north–south decreasing pattern in mud content(Molinaroli et al.,2007).Water salinity is influenced by freshwater inputs.2.2.Yield dataThe latest available commercial yield data in the Venice lagoon were relative to year2007(Fig.2a)for∼65%of areas in con-cession for exploitation of R.philippinarum(Michele Pellizzato, unpublished data).While there exists a substantial variation in pro-ductivity within an harvesting area,yield information is usually aggregated at the concession level.Therefore,the spatial distribu-tion of clam yield within a concession area was derived by using biomass data of R.philippinarum gathered in a number of indepen-dent studies described in G.R.A.L.(2009,and references therein) as a proxy indicator of yield,as described hereafter.Mean size of areas given in concession in2007ranged from2.5to5.5km2.In concession areas for which total landings in year2007and biomass data were available(n=13,out of the total24concession areas), a relative biomass index was computed by dividing the biomass of each100×100m cell by the total biomass of the area.The yield distribution for each100×100m cell within a concession was then obtained by multiplying the annual production for the area by the relative biomass index.2.3.Environmental factorsAs reported by Paesanti and Pellizzato(2000),R.philippinarum is quite tolerant to mid-high variations of relevant habitat variables typical of coastal lagoons,such as salinity,temperature,dissolved oxygen and turbidity.In the present study,we chose a set of nine environmental variables that are known to affect clam abun-dance and yield(Barillari et al.,1990;Paesanti and Pellizzato,2000; Vincenzi et al.,2006b,2007),namely:percentage of sand in the sediment(Sand),dissolved oxygen(DO),salinity(Sal),water speed (Speed),chlorophyll“a”(Chla),turbidity(Turb),residence time (RT),water temperature(T)and water depth(WD).Several studies,sampling surveys and monitoring programs have been carried out in the Venice lagoon in the lastfive years1474S.Vincenzi et al./Ecological Modelling222 (2011) 1471–1478to gather information on the physical,chemical and hydrolog-ical characteristics of the lagoon.Point values of sand content (n=150)provided by the Venice Water Authority(MAG ACQUE –SELC,2005;MAG ACQUE–Thetis,2005)were interpolated by using ordinary kriging on the grid chosen(100×100m cells),after fitting the best model on the experimental variogram,using the ‘gstat’package(Pebesma,2004)for the software R(R Development Core Team,2009).Data of water quality(n=15)obtained from the Venice Water Authority(MAG ACQUE–SAMA–Thetis,2007)were interpolated by using ordinary kriging to create maps of water temperature,turbidity,salinity,chlorophyll-“a”,and dissolved oxy-gen.Available data for2007were interpolated to generate monthly maps of these environmental factors,then the map of yearly mean values was computed for each factor by averaging monthly data. Hydrodynamism for a typical tidal cycle was acquired from the-matic maps with categorical discretization provided by Molinaroli et al.(2007),while residence time data were given by Cucco et al. (2009).2.4.The calibration and validation datasetsFor2539cells within the harvesting areas we had information about both yield for year2007and the nine environmental variables listed in Section2.2.The dataset was randomly split in a calibration dataset(CD,n=1698)and a validation dataset(VD,n=841).The relationship between yield data and the environmental variables was modelled by using the calibration dataset and the quality of predictions was then assessed by using the validation dataset,as described in Section2.4.2.5.Statistical methodsCollinearity among environmental variables was tested by hier-archical cluster analysis using squared Spearman correlations( 2) as similarity measure.To model the relationship between the nine environmental vari-ables and site-specific yield of R.philippinarum in the Venice lagoon, we used the Random Forest algorithm(RF,Breiman,2001)imple-mented in the“randomForest”package(Liaw and Wiener,2002) within the R environment(R Development Core Team,2009).RF is an ensemble learning technique developed by Breiman (2001)based on a combination of a large set of decision trees.Each tree is trained by selecting a random set of variables and a ran-dom sample from the training dataset(i.e.,the calibration data set). Three training parameters needs to be defined in the Random Forest algorithm:ntree,the number of bootstrap samples for the original data(the default value is500);mtry,the number of different pre-dictors tested at each node(which,in this specific case,can be9at most,i.e.,as many as the environmental covariates);nodesize,the minimal size of the terminal nodes of the trees,below which leaves are not further subdivided.As the response variable(yield of R.philippinarum)was numer-ical,we confine our attention to regression Random Forest models. Among the predictors,only hydrodynamism and residence time entered the model as categorical variables,as they were acquired from thematic maps in Molinaroli et al.(2007)and Cucco et al. (2009).The algorithm performs as follows(for full details see Breiman,2001):(1)ntree bootstrap samples X i(i=bootstrap iteration)are randomlydrawn with replacement from the original dataset(training dataset),each containing approximately two third of the ele-ments of the original dataset X(in our case approximately1132 elements out of1639).The elements not included in X i are referred to as out-of-bag data(OOB)for that bootstrap sample.(2)For each boostrap sample X i an unpruned regression tree isgrown.At each node,rather than choosing the best split among all predictors as done in classic regression trees,mtry variables are randomly selected and the best split is chosen among them.(3)New data(out-of-bag elements)are predicted by averaging thepredictions of the ntree trees,as explained below.Out-of-bag elements are used to estimate an error rate,called the out-of-bag(OOB)estimate of the error rate(ERR OOB),as follows:(i)At each bootstrap iteration,the out-of-bag elements are pre-dicted by the tree grown using the bootstrap samples X i.(ii)For the i th element(y i)of the training dataset X,all the trees are considered in which the i th element is out-of-bag.On average, each element of X is out-of-bag in one-third of ntree iterations.On the basis of the random trees an aggregated prediction g OOB is developed.The out-of-bag estimate of the error rate is com-puted as ERR OOB=(1/ntree)ntreei=1[y i−g OOB(X i)]2.The ERR OOB help prevent overfitting and can also be used to choose an optimal value of ntree and mtry,by selecting ntree and mtry that minimize ERR OOB.Therefore,wefirst chose the optimal values of ntree and mtry which minimize ERR OOB and then we pro-ceeded to develop the Random Forest model.The“randomForest”package can also produce a measure of vari-able importance by looking at the deterioration of the predictive ability of the model when each predictor is replaced in turn by ran-dom noise.The resulting deterioration is a measure of predictor importance.The most widely used score of importance of a given variable in regression RF models is the increasing in mean of the error of a tree(mean square error,MSE).In addition,partial plots provide a way to visualize the marginal effect of environmental variables in Random Forests estimates of potential yield.As ERR OOB is an unbasied estimate of the generalization error,in general it is not necessary to test the predictive ability of the model on an external data set(Breiman,2001).However,we preferred to use an independent dataset(the VD data set with841mea-sures of yield and environmental variables)to perform an external validation of the predictive capabilities of the RF model.2.6.Predictive mapsOnce calibrated and validated,the resulting RF model was applied to the entire lagoon of Venice to obtain an estimate of the potential yield of R.philippinarum for the whole basin.The potential yield of sub-basins and of all harvesting areas given in concession in 2007were also computed(Fig.2a).The RF model wasfinally used also to estimate yield potential of the areas(Fig.2b)where clam harvesting will be allowed starting from2013(G.R.A.L.,2009).RTWDSalChlaTurbSandSpeedTDO 0.8.6.4.2.Spearmanρ2Fig.3.Hierarchical clustering using squared Spearman correlation( 2)of environ-mental variables as similarities.Sand,share of sand in the sediment;Sal,salinity; WD,water depth;DO,dissolved oxygen;Turb,turbidity;RT,residence time;Chla, chlorophyll-“a”;T,water temperature;Speed,water speed.S.Vincenzi et al./Ecological Modelling 222 (2011) 1471–14781475SpeedT Chla RT Turb DO WD Sal Sand ●●●●●●●●●5045403530%IncMSEFig.4.Variable importance plot generated by the random forest algorithm included in the randomForest package for R software.The plot shows the variable importance measured as the increased mean square error (%IncMSE),which represents the dete-rioration of the predictive ability of the model when each predictor is replaced in turn by random noise.Higher %IncMSE indicates greater variable importance.3.ResultsA strong correlation was found between chlorophyll “a ”and turbidity (Fig.3).The out-of-bag estimates of the error rate (ERR OOB )were used to select the optimum Random Forest parameters (mtry =3,ntree =700,nodesize =5).For the calibration dataset (CD),the Ran-dom Forest was able to explain a large proportion of the variance of yield of R.philippinarum (r 2CD =0.99).The out-of-bag validation results were examined,for which r 2OOB =0.93.Fig.4shows the ranking of predictors by their importance.Only few of the descriptors contributed noticeably to the estimation of yield of R.philippinarum ,namely percentage of sand in the sed-iment (Sand),salinity (Sal)and water depth (WD).In decreasing order of importance the other predictors included in the RF model were:dissolved oxygen,turbidity,residence time,chlorophyll-“a ”,temperature and current speed.Partial plots representing the marginal effect of single variables included in the RF model on estimates of yield of R.philippinarum are shown in Fig.5.The Random Forest model provided a good prediction of pro-duction values in the validation dataset (r 2VD =0.74,Fig.6).The results of the RF model tended to overestimate yield in low yield sites (Fig.6).For the whole 216km 2of Venice lagoon (excluding emerged lands,seagrass meadows and areas in which harvesting is forbidden for sanitary reasons)the RF model estimated a poten-tial yield of about 70,000t y −1(Table 1),with an average yield of 321g m −2y −1(Table 1).Maximum yield potential was predicted in the central part of the Central sub-basin,while the Northern sub-basin presented the lowest yield per unit area (Fig.7and Table 1).The estimated yield for the areas harvested in 2007was about 22,700t (Table 1),with an average yield of 567g m −2y −1(Table 1).77%of the total yield of the areas given in concession in 2007was located in Central sub-basin,where only 63%of the surface of the harvesting areas were located (Table 1).The proposed reshape of concessions led to similar level of average yield potential for the Central and the Southern sub-basins,and to an increase of average yield potential in the Northern sub-basin (Table 1).Due to a reduction of harvesting surface areas,total yield will decrease to about 18,300t,with an average yield of 514g m −2y −1(Table 1).4.DiscussionWe showed that the application of a Random Forest model pro-vides an effective methodology for identifying suitable sites and quantifying site-specificyields for the exploitation of an aquacul-ture species.Random Forests,both classifier and regression,haveFig.5.Partial plots representing the marginal effect of single variables included in the RF model on estimates of yield of R.philippinarum while averaging out the effect of all the other variables.In a partial plots of marginal effects,only the range of values (and not the absolute values)can be compared between plots of different variables.1476S.Vincenzi et al./Ecological Modelling 222 (2011) 1471–1478300025002000150010005000050010001500200025003000Observed annual yield (g m −2)Observed annual yield (g m −2)P r e d i c t e d a n n u a l y i e l d (g m −2)30002500200015001000500050010001500200025003000VDCDFig.6.Application of the RF model to the calibration data set (CD,r 2=0.99,p <0.01)and the validation data set (VD,r 2=0.74,p <0.01).Predictions of the RF models are more uncertain in sites with low yieldpotential.Fig.7.Map showing the prediction of yield of R.philippinarum in the Venice lagoon obtained by the application of the Random Forest model.Areas given in concession in 2007and in 2013,according to the reshaping plan proposed by the Province of Venice,are showed.been already used in several applicative context and were recently applied to predict plant and animal habitat suitability (e.g.,Iverson et al.,2005;Lawler et al.,2006,2009;Benito Garzón et al.,2007,2008).The results of the RF model tended to overestimate yield in low yield sites and were more accurate for increasing yields (Fig.6).In order to assess if the overestimation was dependent on the par-ticular subsets of the data used for calibration and validation,weS.Vincenzi et al./Ecological Modelling222 (2011) 1471–14781477re-fitted the RF model using other splits,but the RF model param-eters and predictions did not substantially change.As sites of low yield are obviously sites of low commercial interest,the relative underperformance of the model in those sites does not hinder the application of the model for the identification of suitable sites for clam harvesting and the estimation of site-specific yield potential. However,this could substantially overestimate yield outside con-cession areas,as many of the sites not harvested are not suitable for R.philippinarum(Fig.7).Therefore,the prediction of70,000t for the whole lagoon must be taken with great caution and only rep-resent a gross estimation of potential yield for the whole Venice lagoon.Another important aspect of modelling involves the evaluation and biological interpretation of the results.In the case of R.philip-pinarum,the results are encouraging.The marginal effects of single environmental factors(Fig.5)confirm thefindings of Barillari et al. (1990),Paesanti and Pellizzato(2000)and Vincenzi et al.(2006a,b, 2007)on their optimal values for R.philippinarum growth and sur-vival in North Adriatic lagoons.Share of sand in the sediment play a major role in determining the yield of R.philippinarum in the Venice lagoon(Fig.4).Sev-eral studies showed that greater growth rates,maximum size and successful juvenile settlement in R.philippinarum occur in sandy sediments than in sediments with higher fraction of silt(Barillari et al.,1990;Rossi,1996;Meliàet al.,2004).Moreover,share of sand in the sediment was the most important factor in determining both presence/absence of the species and its abundance also in the habi-tat suitability models developed by Vincenzi et al.(2006a,b,2007) for R.philippinarum in the Sacca di Goro lagoon.The other two most important predictors were salinity and water depth.According to Paesanti and Pellizzato(2000),the opti-mal values for salinity range between25and35PSU.In the RF model,the marginal effect of salinity increased for salinity values greater than30PSU.As for water depth,areas shallower than0.5m are not suitable for R.philippinarum,as clams could emerge with low tide and unfavorable wind conditions and are also vulnerable to predation by birds.As harvesting is carried out by mechanical dredging of the bottom,sites with water depth greater than10m are in general not suitable for commercial exploitation,especially in sites where clams are seeded.Vincenzi et al.(2006a)found that the three most important predictors of yield of R.philippinarum were share of sand in the sediment,salinity and water speed.Surpris-ingly,in the RF model,water speed was the least important factor in terms of marginal effect(Fig.4).This further confirms the neces-sity of site-specific calibrations of correlative models of species abundance.Optimal sites for clam farming are characterized by intermediate currents,typically from0.3to1.5m s−1(Paesanti and Pellizzato,2000).The predictive maps obtained by the application of the RF modelfitted on the calibration data set(CD)showed strikingly site-specific differences in potential yield(Fig.7).Turolla et al. (2008)estimated by using expert knowledge in c.a.27,500t the total production for2007,accounting for unauthorizedfishing,that isfishing occurring in polluted areas(Pellizzato and Da Ros,2005).It is worth noting that no official landing data for the whole lagoon are available.Considering the16%reduction of the surface of exploited area due to environmental(i.e.,presence of seagrass meadows),health-risk constraints and the attempt to maximize production for unit area given in concession,the total production estimated for the new configuration of concessions seems to be adequate with respect to local managers expectations(13,000t,G.R.A.L.,2006).However, some areas given in concession in2007and also maintained in the2013plan,mostly in the Northern and Southern sub-basins, showed low potential yield,according to model predictions.It is clear that model predictions for areas which will be given in concession in2013did not take into account possible changes in environmental conditions in the lagoons,both natural and human-induced,that might alter the distribution of suitable sites within the lagoon,and further restrictions on exploitation activities(e.g.,fishing days,type of harvesting tools,individual quotas,etc.).In addition,as the RF model wasfitted on data from a single year (2007),additional studies should be carried out to investigate the relative importance of annualfluctuations of the biogeochemical and hydrodynamic factors included in the RF model in determin-ing the observed inter-annual variability of the commercial yield of R.philippinarum in the Venice lagoon(Pellizzato and Da Ros,2005) and to assess if the inclusion of other biogeochemical parameters could further improve the predictions of the RF model.The application of a Random Forest model(both classifier and regression)to predict the distribution(occurrence and abundance) of a species is particular useful when there are complex inter-actions between predictors and response variable(in our case the yield R.philippinarum)and the possibility of highly corre-lated predictor variables.A further advantage of the RF model is that this statistical learning modelling framework does not require assumptions of normality of model variables and can deal with non-linear relationships.Here,the application of a RF model was particularly recommended,as previous models for R.philip-pinarum developed for specific application to the Sacca di Goro lagoon(Italy)clearly showed the non-linearity of the relationship between environmental factors and yield potential(Vincenzi et al., 2006a,2007)and at least two environmental variables were highly correlated.In the context of aquaculture,correlative approaches,in which the relationship between the presence or abundance of a species and environmental conditions is statistically analyzed,are most useful when spatially-explicit information on the occurrence or abundance of the investigated species is available and a measure of the suitability of sites for harvesting is the most important result to be obtained from the analysis.Both mechanistic and correlative approaches are currently used to model species distribution(see Buckley et al.,2010for a recent review).The goal of the correlative RF model presented here was the identification of areas with different degree of suitabil-ity for clam farming and the corresponding yield potential under the assumption that the well-established day-to-day management and rearing practices are carried on.On the contrary,mechanistic approaches were used by Pastres et al.(2001),Solidoro et al.(2003), Meliàet al.(2003,2004)and Spillman et al.(2008)to analyse opti-mal management strategies or to identify suitable rearing sites for R.philippinarum.These models,based on functional traits and phys-iological constraints and making use of complex3D models of water circulation,although costly to design,calibrate and validate,are particularly appropriate to address issues such as long-term sus-tainability of exploitation activities,effects of alternative rearing strategies(e.g.,seeding size and density),risk of dystrophic crises and algal blooms.For instance,in the Venice lagoon an enormous amount of seed is needed each year(7billion individuals,Pellizzato and Da Ros,2005),and while seed is produced naturally in high abundance,densities of juveniles are greatest in areas characterized by high organic pollution.Thus,great attention must be devoted to the implementation of efficient rearing strategies and mechanis-tic models can be a valuable tool for the evaluation of alternative strategies.In this case,our RF model could:(i)provide informa-tion on site-specific yield potential,especially for areas outside the concession areas and thus less investigated with dynamic models; (ii)guide the application of the mechanistic model,for instance by limiting the costly application of computing-intensive mechanistic models to sites where yield potential is above a certain thresh-old;and(iii)suggest particular traits or processes to include in a mechanistic model.。
类普鲁士蓝的制备及其活化PMS降解双酚S
化工进展Chemical Industry and Engineering Progress2023 年第 42 卷第 12 期类普鲁士蓝的制备及其活化PMS 降解双酚S杨有威1，2，3，曾亦婷1，2，郭昌胜3，罗玉霞1，2，高艳1，2，王春英1，2（1 矿冶环境污染防控江西省重点实验室，江西 赣州 341000；2 江西理工大学资源与环境工程学院, 江西 赣州341000；3 中国环境科学研究院环境基准与风险评估国家重点实验室, 北京 100012）摘要：通过简单共沉淀法合成了类普鲁士蓝化合物（CoFe-PBA ），用于活化过一硫酸盐（PMS ）降解有机污染物双酚S （BPS ）。
使用扫描电镜、X 射线衍射、X 射线光电子能谱等手段对CoFe-PBA 进行表征，结果表明CoFe-PBA 由紧密结合的Co 3[Fe(CN)6]2构成，为纳米级，表面均匀分布着C 、Fe 、Co 、O 元素，具有丰富的活性位点。
催化剂投加量300mg/L 、PMS 投加量400mg/L 、pH=5.89条件下，CoFe-PBA/PMS 降解体系40min 内去除73.77%的BPS ，对酸性和共存离子（SO 42−、NO 3−和Cl −）敏感，碱性环境能促进PMS 快速活化，重复实验显示该体系具有良好稳定性，使用4次后仅下降26.70%，活化性能优于其他材料。
机理分析表明，CoFe-PBA 与PMS 相互作用，作用过程中改变了金属位点价态，发生电子转移，产生各种活性物质降解BPS ，其主要作用活性物种为1O 2；产物分析表明，在CoFe-PBA 活化PMS 系统中，BPS 可历经三种路径最终转化为开环产物及CO 2和H 2O 。
本研究通过低耗能、低成本、快速简易的方法制备CoFe-PBA ，可为活化PMS 绿色降解BPS 提供思路。
关键词：双酚S ；过硫酸盐活化；类普鲁士蓝；过渡金属中图分类号：X703 文献标志码：A 文章编号：1000-6613（2023）12-6676-11Preparation of Prussian blue and its activation of PMS fordegrading bisphenol SYANG Youwei 1，2，3，4，ZENG Yiting 1，2，GUO Changsheng 3，LUO Yuxia 1，2，GAO Yan 1，2，WANG Chunying 1，2(1 Jiangxi Provincial Key Laboratory of Environmental Pollution Prevention and Control in Mining and Metallurgy,Ganzhou 341000, Jiangxi, China; 2 School of Resources and Environmental Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, Jiangxi, China; 3 State Key Laboratory of Environmental Criteria and Risk Assessment,Chinese Research Academy of Environmental Sciences, Beijing 100012, China)Abstract: A Prussian blue-like compound (CoFe-PBA) was synthesized by a simple co -precipitation method for activating permonosulfate (PMS) to degrade organic pollutant bisphenol S (BPS). CoFe-PBA showed high activity on the removal of bisphenol S from activated PMS. Scanning electron microscopy, X-ray diffraction, and X-ray photoelectron spectroscopy were used to characterize CoFe-PBA. The results showed that CoFe-PBA was composed of Co 3[Fe(CN)6]2, which was in nanometer scale. The surface was evenly distributed with C, Fe, Co, O elements, with abundant active sites. Under the conditions of catalyst研究开发DOI ：10.16085/j.issn.1000-6613.2023-0803收稿日期：2023-05-15；修改稿日期：2023-07-21。
亚洲区域陆地生态系统碳通量观测研究进展
七、结论
总的来说，亚洲区域的陆地生态系统碳通量观测研究在设备研发、观测技术 改进以及典型生态系统的CO2通量特征和环境控制机理等方面取得了一些重要进 展。然而，仍面临一些科学问题需要解决。未来，需要进一步加强新技术的研究 和应用，深入探索气候变化和人类活动对生态系统碳循环的影响机制，并积极参 与国际合作，共同推动亚洲地区的陆地生态系统碳通量观测研究的发展。
亚洲区域陆地生态系统碳通量 观测研究进展
目录
01 一、引言
三、典型生态系统
03 CO2通量特征及环境 控制机理
二、通量观测设备的
02 研发与观测技术的改 进
04
四、生态系统碳循环 模型模拟
目录
05 五、面临的科学问题 与解决途径
07 七、结论
06
六、新洲地区拥有丰富的生态环境和独特的生态系统，包括森林、草原、沙漠和 湿地等。这些生态系统在全球碳循环中起着重要的作用，它们不仅吸收大量的二 氧化碳（CO2），还通过呼吸作用释放出大量的碳。因此，对亚洲区域陆地生态 系统碳通量的观测和研究，对于理解全球碳循环过程，预测气候变化趋势，以及 制定相应的环境政策具有重大意义。
针对这些问题，解决途径主要包括：一是加强新技术的研究和应用，如无人 机观测和卫星遥感等，以提高数据采集的效率和精度；二是深入研究气候变化和 人类活动对生态系统碳循环的影响机制，为制定相应的环境政策提供科学依据； 三是加强国际合作，通过共享数据和资源，共同解决面临的科学问题。
六、新的区域合作机会
亚洲地区的陆地生态系统碳通量观测研究不仅需要本地区的科学家们的努力， 也需要国际社会的支持和合作。例如，可以通过参与国际研究项目，如 "FLUXNET"等，来提高亚洲地区的研究水平和技术能力。同时，也可以通过举办 国际会议和研讨会等方式，加强学术交流和合作。
中国草原生态价值及时空动态格局
第 32 卷第 12 期Vol.32，No.121-132023 年 12 月草业学报ACTA PRATACULTURAE SINICA李佳慧，黄麟，樊江文. 中国草原生态价值及时空动态格局. 草业学报， 2023， 32（12）： 1−13.LI Jia-hui，HUANG Lin，FAN Jiang-wen. Ecological value and its spatiotemporal dynamic patterns of grassland in China. Acta Prataculturae Sinica，2023， 32（12）： 1−13.中国草原生态价值及时空动态格局李佳慧1，2，黄麟3*，樊江文3（1.中国科学院地理科学与资源研究所，资源与环境信息系统国家重点实验室，北京100101；2.中国科学院大学，北京 100049；3.中国科学院地理科学与资源研究所，陆地表层格局与模拟重点实验室，北京100101）摘要：草原是我国最大的陆地生态系统、重要的自然资源、人-草-畜社会生态系统的载体，然而仍有大面积草原存在不同程度的退化，亟待保护修复以提升质量、功能和稳定性。
本研究利用生态价值核算作为一种监测和评估草原多重生态功能变化的有效途径，更新并分析了2000-2020年中国草原生态系统功能及其价值的地域分异特征，评价了草原核心生态价值的时空演变态势，并基于核心生态功能及价值变化方向和程度提出了分区分类的草原保护修复优化提升策略。
结果表明：1）2020年，中国草原潜在生态价值约24.7万亿元，每km2约760万元，以防风固沙（27.3%）和物种保育（25.8%）为主。
2）近20年，超过90%草原的生态价值呈增长趋势，特别是青藏高原东部、黄土高原北部和内蒙古中部东部等。
3）省域比较而言，蒙、藏、青、川、新的草原生态价值之和约占全国的67.4%，近20年增长较多的省域为陕西、北京、宁夏、天津与山西，增幅均超过65%。
基于MODIS数据的洞庭湖生态经济区生态环境质量演变研究
future development planning. The research results can provide an important theoretical basis for the sustainable development of the area around Dongting Lake.Key words Dongting Lake ecological economic zone; Google Earth Engine; ecological quality; remote sensing ecological ind ex生态环境质量是生态系统在时间和空间上的要素、结构及功能综合表征，反映各种限制因素、景观要素和生态水文过程相互作用的结果[1]。
因此，对于区域生态环境的正确认识及评价对我国生态文化建设和生态环境维护意义重大。
卫星遥感技术能快速覆盖大面积区域并及时获取地表信息，已被广泛应用于生态环境评估领域 [2， 3]。
各种各样的遥感指数已被广泛应用于森林、草地、城市和湖泊等生态环境的评估中[4-7]，如归一化植被指数（NDVI）、增强植被指数（EVI）、叶面积指数（LAI）等[8-10]，这些单一遥感指数是地域生态环境评估的关键之一。
然而，由于生态指标影响因素的复杂性和多样性，仅采用一种遥感指数来量化生态系统的状况是不够的[11]，使用多个指标的综合生态指数来量化评估生态状况更具优势，也更加全面。
早在2006年，生态环境部根据生物丰度指数、植被覆盖指数、水网密度指数、土地退化指数和环境质量指数构建的生态环境状态指数（EI），在区域生态环境质量评估方面取得了普遍应用[12]。
但是，综合指数构建普遍面临评价指标提取困难、数据空间精度较低和数据更新慢等问题。
基于卫星遥感信息结合了绿度、湿度、热度、干度的遥感生态指数（RSEI），可以较好地解决上述问题[13]。
RSEI 的指标易于获取且计算简捷，无须人为设定权重和摘要 快速城镇化对地区生态环境质量产生重要影响，及时评估生态环境质量变化对城市生态管理和规划具有重要意义。
近百年全球草地生态系统净初级生产力时空动态对气候变化的响应
第25卷第11期Vol.25,No. 11草业学报八CTA PR八丁八CULTUR八E SINIC八1-14 2016年11月DOI：10. 11686/cyxb2016148 http： //cyxb. lzu. edu. cn 刚成诚，王钊齐，杨悦，陈奕兆，张艳珍，李建龙，程积民.近百年全球草地生态系统净初级生产力时空动态对气候变化的响应.草业学报，2016, 25(11) ：1-14.GANG Cheng-Cheng, WANG Zhao-Qi, YANG Yue, CHEN Yi-Zhao, ZHANG Yan-Zhen, LIJian-Long, CHENG Ji-Min. The NPP spatiotemporal variation of global grassland ecosystems in response to climate change over the past 100 years. Acta Prataculturae Sinica» 2016，25(11) ：1-14.近百年全球草地生态系统净初级生产力时空动态对气候变化的响应刚成诚u，3%王钊齐3，杨悦3，陈奕兆3，张艳珍3，李建龙3%程积民〃(1.西北农林科技大学水土保持研究所，陕西杨凌712100;2.中国科学院水利部水土保持研究所，陕西杨凌712100 ;3.南京大学生命科学学院，江苏南京210093)摘要：气候变化是影响生态系统空间地理分布、结构和功能的主要因素。
为了从长时间序列大空间尺度上了解气候变化对草地生态系统的影响及其反馈机制，本研究利用综合顺序分类法及分段模型分别模拟了 1911 一2010年间全球草地生态系统及净初级生产力（NPP)的时空动态，并通过相关性分析揭示草地N PP对不同气候因子的响应。
结果表明，在过去的百年间，全球草地面积从1920s的51乃.73万km2下降到1990s的5102. 16万km2，其中冻原与高山草地类组的面积下降最多，为192.35万km2，荒漠草地类组、典型草地类组和温带湿润草地类组的面积分别下降了 14. 31、34. 15和70. 81万km2，而热带萨王纳类组的面积增加了 238. 06万km2。
升金湖和菜子湖越冬白额雁肠道寄生线虫多样性研究
第41卷第6期生态科学41(6): 202–210 2022年11月Ecological Science Nov. 2022 李清月, 龚治忠, 冯佳慧, 等. 升金湖和菜子湖越冬白额雁肠道寄生线虫多样性研究[J]. 生态科学, 2022, 41(6): 202–210.LI Qingyue, GONG Zhizhong, FENG Jiahui, et al. Intestinal parasitic nematode diversity in communities of Wintering Greater White-Fronted Geese (Anser albifrons) between Caizi Lake and Shengjin Lake in China[J]. Ecological Science, 2022, 41(6): 202–210.升金湖和菜子湖越冬白额雁肠道寄生线虫多样性研究李清月, 龚治忠, 冯佳慧, 刘刚*安徽医科大学生命科学学院, 合肥 230032【摘要】肠道寄生虫是鸟类中常见的病原体之一, 可以传播给人类及其他动物, 引起严重的人禽共患病。
通过采集菜子湖和升金湖60份越冬白额雁粪便样本, 以肠道寄生线虫特异引物作为标记基因进行高通量测序, 分析两个湖泊越冬白额雁的肠道寄生线虫感染种类、多样性及其群落组成结构。
将3299430条reads归属为肠道寄生线虫门, 共定义3249个肠道寄生线虫的OTUs, 鉴定到4个目, 91属和191种; 4个目分别为小杆目(Rhabditida)(62.12 %)、垫刃目(Tylenchida)(32.20%)、疏毛目(Araeolaimida)(3.40%)和单宫目(Monhysterida)(2.27%), 其中小杆目为优势物种。
Shannon-wiener指数和Simpson指数显示, 菜子湖越冬白额雁的肠道寄生线虫alpha多样性高于升金湖越冬白额雁, 而差异性不显著, PCA聚类分析表明, 两个湖泊中越冬白额雁肠道寄生线虫群落的组成结构相似。
广东省典型作物轮作优化组配专家系统
第41卷第6期生态科学41(6): 73–81 2022年11月Ecological Science Nov. 2022 叶延琼, 王悦, 章家恩, 等. 广东省典型作物轮作优化组配专家系统[J]. 生态科学, 2022, 41(6): 73–81.YE Yanqiong, WANG Yue, ZHANG Jia’en, et al. Typical Crop Rotation Optimization Expert System in Guangdong province[J]. Ecological Science, 2022, 41(6): 73–81.广东省典型作物轮作优化组配专家系统叶延琼1,2,3,4, 王悦1, 章家恩1,2,3,4,*, 孔旭晖1,2,3,4, 秦钟1,2,3,41. 华南农业大学资源环境学院, 广州 5106422. 广东省生态循环农业重点实验室, 广州 5106423. 广东省现代生态农业与循环农业工程技术研究中心, 广州 5106424. 农业部华南热带农业环境重点实验室, 广州 510642【摘要】基于广东省60余种典型作物的生长属性、经济效益等基础数据, 借助Visual Studio 2012运行平台, 使用C sharp编程语言结合Access数据库, 开发构建了广东省典型作物轮作优化组配专家系统V1.0(中国计算机软件著作权登记号: 2018SR1064279)。
此系统可以根据作物的生长属性, 筛选出某地适宜种植的作物库, 并根据优选的各种作物的生育期进行优化组合, 输出可选的作物轮作组配方案; 在此基础上, 根据作物的经济效益及相关特定判断规则, 输出适宜某地区作物轮作的优化组配方案, 进而为生产者提供作物轮作模式的多元及优化选择。
关键词：专家系统; 作物轮作; 优化; 广东省doi:10.14108/ki.1008-8873.2022.06.009 中图分类号：S126 文献标识码：A 文章编号：1008-8873(2022)06-073-09Typical Crop Rotation Optimization Expert System in Guangdong provinceYE Yanqiong1,2,3,4, WANG Yue1, ZHANG Jia’en1,2,3,4,*, KONG Xuhui1,2,3,4, QIN Zhong1,2,3,41. College of Natural Resources and Environment, South China Agricultural University, Guangzhou 510642, China2. Guangdong Provincial Key Laboratory of Eco–Circular Agriculture, Guangzhou 510642, China3. Guangdong Engineering Research Center for Modern Eco-agriculture and Circular Agriculture, Guangzhou 510642, China4. Key Laboratory of Agro-Environment in the Tropics, Ministry of Agriculture and Rural Affairs, Guangzhou 510642, ChinaAbstract: Based on the basic data of over 60 typical crops’ growth attributes and economic benefits in Guangdong province, the “Guangdong Province's Typical Crop Rotation Optimization Expert System V1.0” was established by using the Visual Studio 2012 platform and the C sharp language programming combined with Access database. The Expert System can output the list of appropriate crops to be planted in the specific areas of Guangdong province according to the crop growth attributes; then it can further combine growing season of the recommended crops to provide crop rotation matching schemes for the appropriate crops in the corresponding areas; after that, it can also present a final optimal crop rotation plan by considering with better economical incomes for the specific area according to the crop market price database. This Expert System can provide diversified and optimal options of crop rotation mode for producers and related decision-makers.Key words:Expert System, crop rotation, optimization, Guangdong province收稿日期: 2021-03-26; 修订日期: 2021-05-09基金项目:广东省现代农业产业技术体系建设项目(2018LM1100, 2019KJ105); 广东省科技计划项目(2016A020210094, 2017A090905030); 广东省惠城区丝苗米产业园、连山县丝苗米产业园及罗定市丝苗米产业园等科技支撑服务项目作者简介:叶延琼(1976—), 女, 博士, 副教授, 主要从事GIS、生态服务与生态规划方面的教学与研究工作74 生态科学41卷0 前言作物轮作是指在同一地块上有顺序地轮换种植不同作物或轮换采用不同复种方式的种植方式[1], 按种植方式主要分为旱地轮作(如豆麦轮作、粮棉轮作等)和水旱轮作(如稻菜轮作、稻烟轮作等)[2-4]。
不同石漠化草地根系对土壤有机碳的贡献
第41卷第6期生态科学41(6): 26–32 2022年11月Ecological Science Nov. 2022 张乾, 汪依妮, 柳鑫, 等. 不同石漠化草地根系对土壤有机碳的贡献[J]. 生态科学, 2022, 41(6): 26–32.ZHANG Qian, WANG Yini, LIU Xin, et al. The contribution of roots to soil organic carbon in different rocky desertification grasslands[J]. Ecological Science, 2022, 41(6): 26–32.不同石漠化草地根系对土壤有机碳的贡献张乾1, 汪依妮2, 柳鑫3, 田思惠3, 赵学春1,*1. 贵州大学动物科学学院, 贵阳 5500252. 贵州省草地技术试验推广站, 贵阳 5500253. 中国科学院植物研究所, 北京 100093【摘要】根系是草地生态系统土壤有机碳库的主要供给者。
以潜在、中度和强度石漠化草地群落为研究对象, 采用连续土钻取样法、土柱法和分解袋法, 研究不同石漠化草地根系和土壤有机碳的垂直分布、季节动态、有机碳储量及与土壤因子的关系。
结果表明: 潜在、中度和强度石漠化草地的根系生物量差异显著, 分别为3 355.65 g·m-2、2 944.02 g·m-2和1 806.80 g·m-2。
土壤有机碳含量表现为强度石漠化草地>中度石漠化草地>潜在石漠化草地。
根系和土壤有机碳均趋于土壤表层分布, 0—10 cm土层根系生物量占总根系生物量的57.66%—81.02%, 0—10 cm土层土壤有机碳占总有机碳含量的43.00%—65.60%。
潜在、中度和强度石漠化草地的土壤有机碳储量分别为3.48 Mg、3.93 Mg和3.32 Mg, 通过根系分解补充到土壤的有机碳分别为40.69 g·m-2·a-1、154.79 g·m-2·a-1、57.31 g·m-2·a-1, 占土壤有机碳储量的1.17%、3.94%、1.73%。
气候变暖背景下柴达木盆地生态环境质量遥感监测
第41卷第6期生态科学41(6): 92–99 2022年11月Ecological Science Nov. 2022 李倩琳, 沙占江. 气候变暖背景下柴达木盆地生态环境质量遥感监测[J]. 生态科学, 2022, 41(6): 92–99.LI Qianlin, SHA Zhanjiang. Remote sensing monitoring of ecological environment quality in Qaidam Basin under the background of climate warming[J]. Ecological Science, 2022, 41(6): 92–99.气候变暖背景下柴达木盆地生态环境质量遥感监测李倩琳1, 沙占江1,2,3,4,*1. 青海师范大学, 地理科学学院, 西宁 8100082. 青海省自然地理与环境过程重点实验室, 西宁 8100083. 青藏高原地表过程与生态保育教育部重点实验室, 西宁 8100084. 高原科学与可持续发展研究院, 西宁 810016【摘要】柴达木盆地为典型的高寒荒漠区, 生态环境脆弱, 快速全面地了解其在气候变暖背景下生态环境质量变化具有重要意义。
以2000年、2010年和2020年Landsat TM/OLI遥感影像为数据源, 提取绿度、湿度、干度、热度和盐度作为评价指标, 在主成分分析法的基础上, 提出了柴达木盆地生态环境质量评价方法, 并对其时空变化规律进行了探讨。
结果表明: (1)柴达木盆地生态环境质量整体较为脆弱, 区域差异明显, 呈东南优西北差的分布格局, 自东南向西北环状递减; (2)2000—2020年间, 柴达木盆地生态质量总体呈现改善的趋势, 遥感生态指数均值由2000年的0.330上升到2020年的0.383; (3)生态环境质量改善、退化的区域占比分别为23.97%和5.81%, 改善的地区主要分布在盆地东部、东北部和西部的山地, 退化的地区主要分布在盆地南侧的昆仑山, 以及盆地内部的都兰—诺木洪—格尔木—乌图美仁一线冲洪击扇前缘的绿洲核心区, 盆地内部的沙漠戈壁和盐碱地变化不明显。
ecological processes稿件模板
Ecological Processes稿件模板引言概述:生态学过程是生态学研究的核心内容之一，涉及生态系统中各种生物和非生物要素之间的相互作用和动态平衡。
本文将深入探讨生态过程的关键方面，着重介绍与该领域相关的最新研究成果和理论进展。
通过这一稿件模板，我们旨在为研究者提供一个系统的组织框架，帮助他们更好地撰写与生态过程相关的学术论文。
正文:1. 生态过程的基本概念:1.1 生态学过程的定义1.2 生态过程的分类与特征1.3 生态过程的时空尺度2. 生物多样性与生态过程:2.1 物种多样性对生态过程的影响2.2 生态过程维持和促进生物多样性的机制2.3 损失生物多样性对生态过程的潜在影响3. 气候变化与生态过程:3.1 气候变化对生态过程的直接影响3.2 生态过程对气候变化的反馈效应3.3 气候变化对生态过程的适应性调控4. 生态过程与生态系统稳定性:4.1 生态过程在维持生态系统稳定性中的角色4.2 生态系统退化与生态过程的研究关联4.3 恢复生态过程对生态系统的重建效果5. 人类活动与生态过程的交互:5.1 城市化对生态过程的影响5.2 农业实践与生态过程的关系5.3 生态过程对人类社会的服务功能与回馈总结:通过对生态过程的基本概念、与生物多样性、气候变化、生态系统稳定性以及人类活动的关联进行深入剖析，我们可以清晰地了解生态学过程的多层次、多方面的复杂性。
生态过程既是生态系统运作的引擎，也是人类社会可持续发展的基础。
在未来的研究中，我们需要深化对生态过程的理解，探索更多未知领域，特别是在应对气候变化和人类活动影响方面的研究。
通过共同努力，生态学过程的研究将为我们提供更多关于地球生态系统运行规律的深刻认识，为可持续发展提供更为科学的依据。
这一稿件模板旨在为相关研究者提供一个清晰、系统的写作框架，促进对生态过程的深入研究和讨论。
正文续:3. 生态过程与土壤生态学:3.1 土壤生态过程的关键要素：土壤是生态过程中一个至关重要的组成部分，其生态学过程包括土壤微生物的活动、有机物质的分解、以及养分的循环。
城市绿地中鸟类对海南蒲桃的取食和传播作用
第41卷第6期生态科学41(6): 100–104 2022年11月Ecological Science Nov. 2022 汪国海, 董佩佩, 韦丽娟, 等. 城市绿地中鸟类对海南蒲桃的取食和传播作用[J]. 生态科学, 2022, 41(6): 100–104.WANG Guohai, DONG Peipei, WEI Lijuan, et al. Fruit foraging and dispersal of Syzygium cumini by frugivorous birds in urban green space[J]. Ecological Science, 2022, 41(6): 100–104.城市绿地中鸟类对海南蒲桃的取食和传播作用汪国海1, 董佩佩1, 韦丽娟2, 黄秋婵1, 韩巧1, 唐创斌1,*1. 广西民族师范学院化学与生物工程学院, 广西崇左 5322002. 广西民族师范学院数理与电子信息工程学院广西崇左 532200【摘要】海南蒲桃(Syzygium cumini)是亚热带地区城市绿地中常见的绿化植物, 其果实数量多、果期长, 可为鸟类提供大量食物资源。
2020年6月—8月借助 Safari 10×26 变焦双筒望远镜, 采用焦点扫描法对访问海南蒲桃果实(种子)的鸟类行为进行观察, 详细记录鸟类的种类、取食基质、访问频次、取食时间、取食数量和取食方式等信息, 探讨鸟类在海南蒲桃种子传播及种群更新中的生态作用。
结果表明: 成熟的海南蒲桃能吸引7种食果鸟类对其种子进行取食,其中白头鹎(Pycnonotus sinensis)、红耳鹎(Pycnonotus jocosus)、白喉红臀鹎(Pycnonotus aurigaster)和黄臀鹎(Pycnonotus xanthorrhous)4种鸟类以整吞的方式取食海南蒲桃的种子, 属于种子潜在传播者。
不同种鸟类对海南蒲桃果实的取食频次间存在显著差异(t=4.310, df=6,P < 0.01), 平均访问只数、平均取食时间和平均取食量间存在极显著差异(P < 0.001)。
国家中心城市交通运输业碳排放效率研究
第41卷 第1期 生 态 科 学 41(1): 169–1782022年1月 Ecological Science Jan. 2022收稿日期: 2020-06-05; 修订日期: 2020-08-01基金项目: 广东省自然科学基金项目(2015A030310413); 广州市属高校科研项目(1201431115)作者简介: 任梦洋(1997—), 女, 河南周口人, 硕士研究生, 主要从事环境管理研究,E-mail:*****************通信作者: 黄羿, 女, 博士, 讲师, 主要从事环境资源经济学研究,E-mail:****************.cn任梦洋, 黄羿, 付善明, 等. 国家中心城市交通运输业碳排放效率研究[J]. 生态科学, 2022, 41(1): 169–178.REN Mengyang, HUANG Yi, FU Shanming, et al. The study on carbon emission efficiency of transportation industry in national central city[J]. Ecological Science, 2022, 41(1): 169–178.国家中心城市交通运输业碳排放效率研究任梦洋, 黄羿*, 付善明, 常向阳广州大学环境科学与工程学院, 广州 510006【摘要】为改善交通运输业碳排放效率促进城市低碳经济发展, 以在区域发展中发挥引领作用的九个国家中心城市为研究对象, 运用包含非期望产出的Super-SBM 模型和Malmquist-Luenberger 生产率指数, 从静态和动态两个角度对研究区2005—2016年交通运输业碳排放效率进行了测算, 并进一步分析了碳排放效率变化的影响因素。
结果显示, 各年份国家中心城市交通运输业碳排放效率的平均值均偏低, 变化范围在0.4824—0.7609之间, 仍有较大的提升空间。
营口海岸带土地利用及景观格局35年变化
第41卷第6期生态科学41(6): 41–51 2022年11月Ecological Science Nov. 2022 帅艳民, 曲歌, 邵聪颖, 等. 营口海岸带土地利用及景观格局35年变化[J]. 生态科学, 2022, 41(6): 41–51.SHAI Yanmin, QU Ge, SHAO Congying, et al. Land use and landscape pattern changes of Yingkou Coastal Zone over the past 35 Years[J]. Ecological Science, 2022, 41(6): 41–51.营口海岸带土地利用及景观格局35年变化帅艳民1,2,3, 曲歌1,*, 邵聪颖1, 谢东辉4, 朱启疆4, 石莹11. 辽宁工程技术大学, 测绘与地理科学学院, 阜新 1230002. 中国科学院新疆生态与地理研究所, 丝路绿色发展研究中心, 乌鲁木齐 8300113. 中国科学院中亚生态与环境研究中心, 乌鲁木齐 8300114. 北京师范大学, 地理科学学部, 遥感科学国家重点实验室, 北京 100875【摘要】营口沿海产业基地是“兴辽计划”和振兴东北的重要支点, 其经济发展与海岸带变迁的耦合机制亟需海岸带土地利用和景观格局的变化足迹, 以便深入理解海岸城市与经济发展协同演进过程。
以面向对象的分类器为主辅以人工解译, 提取营口海岸带1984—2018年间30m土地利用信息, 采用土地转移矩阵和景观格局指数分析土地利用和景观格局变化趋势, 通过主成分分析法探讨诱发变化的关键驱动因素。
结果表明: (1) 土地利用类型1984—1992年以耕地和盐田变化为主发展到1993—2006年建设用地激增的趋势, 具有从农耕为主到城市化的鲜明时代特征; (2) 土地利用类型间的逐年转化程度各不相同, 其中1993—1995年、2006—2007年、2011—2015年耕地和低效盐田向建设用地的转入最为突出; (3) 景观格局总体上斑块趋于分散, 2004年后破碎化加剧, 并呈现以建设用地和耕地为区域基质, 其他景观周围分布的模式; (4) 人口和经济政策因素在营口市海岸带土地利用方式转变过程中具有明显驱动效应。
广东省不同地貌形态类型区生境质量归因
第41卷 第3期 生 态 科 学 41(3): 24–322022年5月 Ecological Science May 2022收稿日期: 2021-08-12; 修订日期: 2021-11-25 基金项目: 国家自然科学基金项目(42071123)作者简介: 卢茵怡(1998—), 女, 广东江门人, 研究生, 主要研究方向为生态系统服务,E-mail:*****************通信作者: 龚建周(1970—), 女, 湖北恩施人, 博士, 广州大学教授, 主要从事城市生态环境与土地系统评估,E-mail:*****************卢茵怡, 李天翔, 龚建周. 广东省不同地貌形态类型区生境质量归因[J]. 生态科学, 2022, 41(3): 24–32.LU Yinyi, LI Tianxiang, GONG Jianzhou. Attribution of habitat quality in different geomorphological types in Guangdong Province[J]. Ecological Science, 2022, 41(3): 24–32.广东省不同地貌形态类型区生境质量归因卢茵怡1, 李天翔2, 龚建周1,*1. 广州大学地理科学与遥感学院, 广州 5100062. 广州茏腾园林景观设计有限公司, 广州 510520【摘要】城市用地扩张和人类活动生物生境破碎, 已成为生物多样性降低的主要原因; 全面认知区域生境质量是改善生境质量的基础, 更是保护和维护生物多样的前提。
论文基于广东省1980—2018年土地利用变化数据, 利用InVEST 模型对广东省生境质量进行评估, 从地形视角分析其地形梯度效应; 综合自然地理环境和社会经济几个方面, 共选择10个影响因子, 探测不同地形梯度下生境质量的主导因子。
结果表明: (1)林地和耕地是广东省最主要的土地利用类型。
退化泥炭地亚表层土壤酶活性与DOC变化规律研究
第41卷第5期生态科学41(5): 144–151 2022年9月Ecological Science Sep. 2022 曹芹, 刘建亮, 刘坤, 等. 退化泥炭地亚表层土壤酶活性与DOC变化规律研究[J]. 生态科学, 2022, 41(5): 144–151.CAO Qin, LIU Jianliang, LIU Kun, et al. Study on the changes of enzyme activity and DOC in subsoil of degraded peat[J]. Ecological Science, 2022, 41(5): 144–151.退化泥炭地亚表层土壤酶活性与DOC变化规律研究曹芹1, 刘建亮2,3, 刘坤4, 曾嘉1, 严飞1, 杨刚1,*1. 西南科技大学生命科学与工程学院, 绵阳 6210102. 中国科学院成都生物研究所山地生态恢复与生物资源利用重点实验室, 成都 6100413. 中国科学院四川若尔盖湿地生态研究站, 四川红原 6244004. 重庆市生态环境科学研究院固体废物与土壤研究所, 重庆 401147【摘要】在气候变化背景下, 泥炭地亚表层土壤有机碳逐渐参与到碳循环中, 为揭示泥炭地亚表层碳输出与土壤酶活性的关系。
以四川省红原县日干乔湿地自然保护区的泥炭沼泽(S1)、沼泽草甸(S2)、高寒草甸(S3)3种不同退化泥炭生态系统中不同深度(0—30 cm、30—60 cm、60—90 cm、90—120 cm、120—150 cm)的土壤作为研究对象, 研究泥炭地表层(<30 cm)、亚表层(30—60 cm)和深层(>60 cm)土壤酶(酚氧化酶、β-葡萄糖苷酶、蔗糖酶)活性和土壤溶解性有机碳(DOC)的变化规律及二者的关系。
结果显示, 从泥炭沼泽到沼泽草甸再到高寒草甸的退化过程中, DOC含量逐渐增加。
随着泥炭地退化程度的加深, 土壤酶活性呈先升高后降低的变化趋势。
基于犹豫模糊语言的森林火灾发生风险评估
February 2020No. 12020年2月 第1期林业资源管理FOREST RESOURCES MANAGEMENT基于犹豫模糊语言的森林火灾发生风险评估骆相宇，李宗敏(四川大学商学院，成都610041)摘要:森林火灾具有分布广、频率高、不确定性强的特点，对森林火灾发生风险进行评估可以为有效防治森林火灾发生、降低损失提供参考。
从气象、地形、可燃物3个维度构建森林火灾发生风险的评价指标体系，首次引 入犹豫模糊语言描述，提高专家语言评价表达的灵活性。
通过犹豫模糊语言混合加权集结算子对语言信息进行集结，得到某区域火灾发生风险的可能性。
该方法被应用到大兴安岭森林某片过火地区，对该地区森林火灾发生风险进行评估，证明了该方法的可行性和有效性。
关键词：森林火灾风险;犹豫模糊语言;犹豫模糊语言混合加权算子中图分类号:S726文献标识码:A 文章编号：1002 - 6622(2020)01 -0183 - 08DOI : 10. 13466/j. cnki. lyzygl. 2020.01.024Risk Assessment of Forest Fire Based on Hesitant Fuzzy LinguisticLUO Xiangyu , LI Zongmin(Business School , Sichuan University , Chengdu 610041)Abstract : Forest fires have the characteristics of wide distribution , high frequency and strong uncertain ty. The assessment of forest fire risk can provide references for effective prevention and control of forestfires and reduction of losses. The evaluation index system of forest fire risk is constructed from three di mensions of meteorology , terrain and combustibles. It is the first time to introduce hesitant fuzzy linguisticdescription , which improves the flexibility of expert language evaluation expression. Through the hesitant fuzzy linguistic hybrid weighed aggregation operator , to aggregate the linguistic information , then get theprobability of fire risk in a certain area. This method has been applied to an over fire area of Daxinganling forest to assess the risk of forest fire in this area , and this case proves the feasibility and effectiveness ofthis method.Key words : forest fire risk , hesitant fuzzy linguistic , hesitant fuzzy linguistic hybrid weighted aggregation operators森林作为地球上最大的陆地生态系统,在人类 过光合作用吸收大量的二氧化碳，转化为有机物,生存和发展的历史上起着不可替代的作用。
生态足迹文献
Ecological Modelling 222 (2011) 2939–2944Contents lists available at ScienceDirectEcologicalModellingj o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /e c o l m o d elTowards a 3D National Ecological Footprint GeographyV.Niccolucci a ,∗,A.Galli b ,A.Reed b ,E.Neri c ,M.Wackernagel b ,S.Bastianoni caEcodynamics Group,Dept.of Chemistry,University of Siena,via della Diana 2A,53100Siena,Italy bGlobal Footprint Network,312Clay Street,Suite 300,Oakland,CA 94607,USA cEcodynamics Group,Dept.of Chemistry,University of Siena,via A.Moro 2,53100Siena,Italya r t i c l ei n f oArticle history:Available online 27 May 2011Keywords:Ecological Footprint FlowFootprint size Footprint depth Stocka b s t r a c tIn the last decades several indicators have been proposed to guide decision makers and help manage natural capital.Among such indicators is the Ecological Footprint,a resource accounting tool with a biophysical and thermodynamic basis.In our recent paper (Niccolucci et al.,2009),a three dimensional Ecological Footprint (3D EF)model was proposed to better explain the difference between human demand for natural capital stocks and resource flows.Such 3D EF model has two relevant dimensions:the surface area (or Footprint size –EF size )and the height (or Footprint depth –EF depth ).EF size accounts for the human appropriation of the annual income from natural capital while EF depth accounts for the depletion of stocks of natural capital and/or the accumulation of stocks of wastes.Building on the 2009Edition of the National Footprint Accounts (NFA),global trends (from 1961to 2006)for both EF size and EF depth were analyzed.EF size doubled from 1961to 1986;after 1986it reached an asymptotic value equal to the Earth’s biocapacity (BC)and remained constant.Conversely,EF depth remained constant at the “natural depth”value until 1986,the year in which global EF first exceeded Earth’s BC.A growing trend was observed after that.Trends in each Footprint land type were also analyzed to better appraise the land type under the higher human induced stress.The usefulness of adopting such 3D EF model in the National Footprint Accounts was also discussed.In comparing any nation’s demand for ecological assets with its own biocapacity in a given year,four hypothetical cases were identified which could serve as the basis for a new Footprint geography based on both size and depth concepts.This 3D EF model could help distinguish between the use of natural capital flows and the depletion of natural capital stocks while maintaining the structure and advantages of the classical Ecological Footprint formulation.© 2011 Elsevier B.V. All rights reserved.1.IntroductionAbout thirty years ago Prof.Enzo Tiezzi,published the first Ital-ian edition of his most famous book Tempi Storici Tempi Biologici (Tiezzi,1984)then translated in English as The end of time (Tiezzi,2003).He wrote:“[...]It is my firm conviction that we must change route as soon as possible and set about defining a new idea of devel-opment.A culture so based will be firmly founded on biology and thermodynamics ,and their fundamental relationship to the econ-omy,the society and the means of production.My conviction is based on three points:(a)the equilibrium of nature is extremely delicate and can be irreversibly upset by man:the resources of nature are not infinite;(b)the destruction of the environment and waste of natural resources is never of long term benefit either economically or socially;(c)the false prosperity of the consumer∗Corresponding author.Tel.:+390577232044;fax:+390577232004.E-mail address:vniccolucci@unisi.it (V.Niccolucci).society is based on the exploitation of three classes of people [...].Dominant economic theory,based as it is on mechanistic princi-ples,remains ignorant of the law of entropy and the role of the time variable.The classical dynamic concept of time and its reversibility,has nothing to do with reality and nature.Time is not without its preferred directions (it is not isotropic)as is space.Time has a direc-tion.Thermodynamics introduces “knowledge of the unidirectional flow of time”,traces the limit between past reality and future uncer-tainty,indicates the orientation of time in natural processes.[...]The technological or economic concept of time is exactly the oppo-site to entropic time.Nature obeys different laws to economics,it works in “entropic time”:the faster we consume natural resources and the energy available in the world,the less time is left for our survival.Technological time is inversely proportional to entropic time,economic time is inversely proportional to biological time .Our limited resources and the limited resistence of our planet and its atmo-sphere clearly indicate that the more we accelerate the energy and matter flow through our Earth system,the shorter is the life span of our species [...].”It is then important to underline “[...]the0304-3800/$–see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.ecolmodel.2011.04.0202940V.Niccolucci et al./Ecological Modelling222 (2011) 2939–2944asymmetry of the ecological and historical time scales:millions of years for the evolution of life on the Earth with extremely slow ecological changes and historical knowledge of only the last brief period(a few thousand years);in contrast to this,the rapid ecolog-ical changes induced by technology in very short historical time. [...]Biological and historical tempos follow different rhythms”.Tiezzi’s thought represents the milestone from which a three dimensional Ecological Footprint(3D EF)model(Niccolucci et al., 2009)was developed.Although the‘classical’Ecological Footprint (EF)method recognizes the crucial role of natural capital and natu-ral income(Rees,2006),it is not sufficiently informative regarding the differentiation between biophysical stocks andflows.Already in the early‘90s,economists belonging to the“thermodynamic school of thought”such as Daly and Georgescu-Roegen(Daly and Farley,2004),widely highlighted the implications of a stock/flow distinction in dealing with sustainability issues.For instance,three operational rules defining the condition of ecological(thermody-namic)sustainability were identified by Daly(1990):(i)renewable resources such asfish,soil,and groundwater must be used no faster than the rate at which they regenerate;(ii)non-renewable resources such as minerals and fossil fuels must be used no faster than their renewable substitutes can be put in place;(iii)pollution and wastes must be emitted no faster than natural systems can absorb,recycle,or render them harmless.Making the role of time explicit within the EF methodology has therefore represented an important step towards a deeper interpre-tation of this indicator in the sustainability framework and could help the EF to better appraise Daly’s rules.As such,the ability to track depletion of natural capital stocks and use of natural capital flows plays a central role within this new approach.The aim of this paper is to test whether the3D EF model recently proposed(Niccolucci et al.,2009)can serve as a useful biophysical measure of theflows and stocks used by a population.To address some of the unanswered questions from the previous paper,global trends for both the size and the depth components of the EF are analyzed here.These trends have been broken down by land type to highlight the areas under critical stress.Finally,the3D EF model is suggested as proxy to redefining a new world geography based on the differentiation betweenflow and stock.2.Ecological Footprint and natural capital accountingHistorically the Ecological Footprint has been presented as a spa-tial indicator for natural capital accounting(Rees,1992).The EF of a population or an individual is defined as the aggregated area of land and water1required on a continuous basis to provide the energy and material resourceflows used and to assimilate the CO2 emissions generated,given prevailing technologies and resource management practices(Wackernagel and Rees,1996).From a ther-modynamic point of view,the Ecological Footprint can be defined as the area continuously required to generate,via photosynthe-sis,a quantity of biomass,and thus negentropy,equivalent to the amount used and dissipated by the population’s consumptive activ-ities(Rees,2006).EF accounts for both direct and indirect land requirements and it is measured in global hectares(g ha)(Wackernagel and Rees,1996; Wackernagel and Kitzes,2008).The term global refers to a normal-ized hectare with world average productivity(Galli et al.,2007; GFN,2009).Land area is a very effective proxy to communicate thefiniteness of planet Earth and its ability to generate resources (Wackernagel and Rees,1997).1Six different land use types are considered by the Ecological Footprint method-ology:croplands,grazing lands,forests,fishing grounds,carbon uptake lands and built-up areas.To give a measure of the(un)sustainability threshold of human consumption,a benchmark called biocapacity(BC)is also provided (Wackernagel and Rees,1996;Monfreda et al.,2004),which quan-tifies how much regenerative capacity exits within a given area, in terms of ecologically productive space.BC is also measured in global hectares(g ha)(Galli et al.,2007;GFN,2009).As EF and BC representflows of ecological assets(i.e.,natural resources and ecological services),they can be directly compared to define an ecological balance in the same way that expenditure and income are compared in economics(Monfreda et al.,2004).The difference between the two terms is proportional to four main fac-tors:(i)population size;(ii)consumption patterns,(iii)ecological productivity and(iv)technology.This ecological balance has significance both at the global and the national level:᭹At global level,the2009Edition of the Global Footprint Net-work’s National Footprint Accounts(GFN,2009)show that the Earth is actually operating in a state of ecological overshoot(EO).Demand for natural resources exceeds the regenerative capac-ity of existing natural capital by44%(GFN,2009;Ewing et al., 2009).Furthermore,the global gap between EF and BC has been continuously increasing since the mid-1980s.From that period up to2002,an ecological debt of about2.5years worth of the Earth’s regenerative capacity has been accumulated as calcu-lated by Kitzes et al.(2008).This debt has likely increased in the last9years and it will keep accumulating until humanity reduces its demand below the Earth’s biocapacity.Though the Earth is characterized by a high resilience,sustained ecological deficit is not possible due to insuperable ecological and thermodynamic constraints.It is thus important to bring our consumption levels back within the limits of our ecological budget.᭹At national level,when the ecological balance is positive (EF<BC),the country analyzed runs an ecological remainder (ER)or surplus.The human load is within the country’s carrying capacity,though this does not necessarily imply sustainable use of domestic ecological resources.This remainder is often used to provide goods and ecological services exported and consumed in other countries,and thus might not constitute an actual remain-der available to the nation.Vice versa,if the balance is negative (EF>BC)then the country is running an ecological deficit(ED), where its natural resource requirements exceed the regenerative capacity of its natural capital.Such an ecological deficit situation also shows the country’s dependence on further goods and eco-logical services,which are provided through either each of the three different mechanisms or a combination of them(Monfreda et al.,2004;Ewing et al.,2010):(a)thefirst is called ecological trade deficit which consists of animport of regenerative capacity from other regions of the world (when possible);(b)the second is known as ecological overshoot.It stimulates anoveruse of resources leading to local and/or global depletion of stocks of natural capital.(c)the third originates from a greenhouse gases accumulation in theatmosphere due to the emission of carbon dioxide faster than the natural absorption rate.Each of these has distinctly different ramifications in terms of local and global sustainability.3.Advances in the Ecological Footprint methodAn advance in the Ecological Footprint method has been pro-posed in our recent paper(Niccolucci et al.,2009),to better explainV.Niccolucci et al./Ecological Modelling 222 (2011) 2939–29442941123024681012142005200019951990198519801975197019651960E F d e p t hE F s i z e (b i l l i o n s g h a )EF depthEFsizeFig.1.The temporal series of absolute EF size (grey line –left side scale)and EF depth (black line –right side scale).Source:our elaboration on Global Footprint Network data.the difference between human demand for stocks and flows via a three dimensional variant of the Ecological Footprint (3D EF).For instance,if the ‘classical’Ecological Footprint methodology (EF classic )can be depicted as a circle,the 3D EF then becomes a cylin-der.In other words,the 3D EF is a volume-based indicator with two relevant dimensions:the surface area (or Footprint size )and the height (or Footprint depth ).The Footprint size (EF size )deals with the human appropriation of the annual income of natural capital provided by the Earth.As sug-gested by Hicks (1946,p.171)the term ‘income’can be considered as the level of consumption which can be sustained in the long run without reducing wealth.In the EF context,the income from natural capital is thus represented by all resource flows and ecological ser-vices annually produced by nature and its biogeochemical cycles.In this sense,EF size deals with the annual appropriation of bioca-pacity and it can assume all values between zero and the annual biocapacity of the planet:0<EF size =BC(1)As EF size is an area,it is expressed in global hectares (g ha)and can be plotted on a (x,y )plane as the basis of the cylinder.The Footprint depth (EF depth )represents the demand for extra land required to meet human needs through depletion of stocks of natural capital and/or saturation of carbon sinks.It can be plotted on the z -axis as it is the height of the cylinder.EF depth can be con-sidered as the number of years necessary to re-generate resources liquidated in one year (and to absorb emitted carbon dioxide)or as the number of planets necessary to support the total consumption of the Earth’s inhabitants.When overshoot (EO)occurs,EF depth is calculated according to Eq.(2),otherwise its value is simply equal to 1:EF depth =1+EO BC(2)EF depth is a dimensionless number that can take any value equal or greater than 1.EF depth ≥1(3)where 1is a reference value termed natural depth.The natural depth corresponds to the intrinsic time (1year)needed by the planet Earth to restore the previous year’s situation by means of its natural flows.When more resources are consumed than are available,an ‘additional depth’is required to accommodate this excess demand.From a sustainability point of view,Footprintdepth should be as close as possible to 1to reduce depletion of stocks.It should be noted that the classical and the three dimensional Ecological Footprint approaches are numerically equivalent and thus the respective final values should be identical.The 3D EF is just a different way of representing classical Footprint values and it is given by the product of the two components:size and depth.EF classic = 3D EF(4)3DEF =EF size ×EF depth(5)4.Results and discussionGlobal EF and BC data drawn from the 2009Edition of the National Footprint Accounts (GFN,2009)were elaborated for the period 1961–2006.Results are reported in Fig.1.Absolute EF size doubles from 1961to 1986.After 1986,it reaches an asymptotic value equal to the Earth’s BC and remains constant until 2006.Conversely,EF depth stays constant at the natural depth value until 1986,the year in which world average EF exceeded Earth’s BC for the first time.Since then a growing trend is observed.In 2006EF depth was equal to 1.44meaning that an extra time of 0.44years (about 5months)would have been necessary to regenerate what humanity consumed in that year.However,this graph could provide misleading information if read in isolation,as it offers a conservative estimation of the real situation.This is due to the fact that when the total Ecological Footprint is calculated by adding up the demand for different land types,an eventual ecological remainder in any given land type is allowed to compensate for ecological deficits in other land types.For example,the deficit of carbon uptake land could partially be compensated by surpluses in grazing land.In other words,the world values for Footprint size and depth (in particular)could be probably higher than those reported in Fig.1.For this reason trends in EF size and EF depth were analyzed for each land type.Results were plotted in Fig.2.Fig.2shows the presence of a depth component just for for-est land.This means that forests are the areas under the highest human induced pressure and that the emission of carbon dioxide is the major driver of global overshoot.A further disaggregation between forest and carbon uptake lands is desirable.Unfortunately,due to data limitations,Global Footprint Network’s National Foot-print Accounts are not able to distinguish between the areas of2942V.Niccolucci et al./Ecological Modelling 222 (2011) 2939–2944123456012345620001990198019701960E F d e p t hE F s i z e (b i l l i o n s g h a )Cropland123456012345620001990198019701960E F d e p t hE F s i z e (b i l l i o n s g h a )Forest123456012345620001990198019701960E F d e p t hE F s i z e (b i l l i o n s g h a )Grazing land123456012345620001990198019701960E F d e p t hE F s i z e (b i l l i o n s g h a )Fishing Ground123456012345620001990198019701960E F d e p t hE F s i z e (b i l l i o n s )Built updepthdepthdepthdepthdepthsizesizesizesizesizeFig.2.The temporal series of EF size (grey line –left side scale)and EF depth (black line –right side scale)for each land type.Forest is given by the sum of forest land and carbon uptake land.Source:our elaboration on Global Footprint Network data.forest dedicated to forest products and those permanently set aside for carbon uptake services (Ewing et al.,2010).The presence of overshoot in forests begins in the mid-1970s,nearly ten years before global overshoot,when both the capacity to produce timber and to uptake CO 2is considered.An imperceptible depth seems to appear also for cropland.Data are not sufficient to assess whether this is a rounding error or the origin of real depth.All other land types report a growing trend in EF size without reaching a plateau (i.e.,biocapacity),while the depth component is fixed on the natural depth value.If the Footprint methodology were to not allow a remainder in a given land type to compensate for deficit in the others,the global EF depth would rise from 1.44to 2.23.The 3D EF approach also enables comparisons between the behavior of different populations,adding more information than classical EF and BC parameters.The comparison among national EF size components can be used as a proxy for the (in)equality in the appropriation of resources and ecological services between current generations of different countries.On the other hand,compar-isons of EF depth values can be used as proxies for the relationships between current and future generations.This new way of representing the Footprint model can be the starting point to create a new Footprint geography based on both size and depth information.Together with the implementation ofa multilateral trade framework in the National Footprint Accounts,the 3D EF model could help us to better track where biocapacity is coming from and where pressures on flows and stocks are taking place.However,when analyzing sub-global systems (i.e.,nations),modifications on Eq.(3)are needed.As nations are open systems able to exchange materials and energy with the surrounding envi-ronments,both the two Footprint components can be considered as the sum of a local and a global term as reported in Eq.(6):3D EFnation =EF size ×EF depth=EF LOC size +EF GLOB size ×EF LOC depth +EF GLOBdepth(6)where:3D EFnation is the total Footprint of consumption of a nation;EF LOCsize and EF GLOB size are the local and global components of the Footprint size of consumption;they refer to the appropriation of annual flows of resources generated inside and outside the given nation,respectively.EF LOC depth and EF GLOBdepth are the local and global components of the Footprint depth of consumption;they refer to the depletion of stocks located inside and outside the given nation,respectively.Please note that the term global refers to all the world other than the single country analyzed.In principle it could be possible to disaggregate this term into small components for each trading part-ner.In practice this can only be done after the implementation of aV.Niccolucci et al./Ecological Modelling 222 (2011) 2939–29442943BAC EF size (gha)BC Nnatural depth=1EF depthD2534Fig.3.Footprint depth vs Footprint size.It is considered a generic nation N with its own Biocapacity (BC N ).Four different cases (named A,B,C,D)can be detected.Case A is characterized by all points laying on the EF size axis from zero to BC N .Case B includes all points on the EF size axis higher than BC N .Both case A and B have a depth value equal to 1(i.e.,natural depth).Case C comprises a set of points where the size component is lower than BC N but the depth component is higher than 1.Case D reports a set of points where the size component is higher than BC N and the depth component is higher than 1.multilateral trade framework into the National Footprint Accounts.Such improvement is expected to be included in the 2012Edition of the Global Footprint Network’s National Footprint Accounts.The implications of this stock/flow distinction could constitute a quantifiable and scientifically sound basis for policy makers to then derive more effective and informed decisions to manage demand on and availability of natural capital,both locally and globally.Interesting information on ecological sustainability of nations can be extrapolated when results are plotted on a EF size vs EF depth plane.Four different hypothetical situations (A,B,C and D)can be identified when comparing a generic nation’s demand for ecolog-ical assets (i.e.,natural resources and ecological services)with its own biocapacity (BC N ),with respect to Eq.(6)(see Fig.3).Case A :nations included in this category consume less ecological assets than those locally available (EF size <BC N )and do not deplete stocks (EF depth =natural depth =1),thus having a long-term repli-cable pathway.A surplus of resource flow for other populations could be potentially available.Although local demand and supply are discussed,this should not lead to the misconception that self-sufficiency is a necessary or desirable criterion for sustainability.There is no physical law or social principle requiring all countries to live within their own biocapacity as countries can access bio-capacity from elsewhere.The only constraint is that while in the short term it is possible for all countries combined to run an eco-logical deficit,this is not possible in the long term as this leads to overshoot and gradual depletion of ecological assets.Theoretically,countries belong to this case if they:(i)have very high biocapacity,which more than compensates the human demand.This is likely the case of European Nordic coun-tries such as Finland and Sweden;and/or(ii)have very low Ecological Footprint with respect to Biocapacity.This is the case of Latin American countries such as Brazil,Peru,and Colombia.Even if all points on the horizontal line from zero to BC N rep-resent favorable conditions from an ecological point of view,this does not ensure that wealth and well being are also met.Additional indicators should be coupled with the Ecological Footprint to con-sider these aspects and draw a more comprehensive picture of the system from a sustainability standpoint.Case B :any country falling into this case consumes more flows than those locally available in a specific year and thus requires addi-tional flows (an “extra size”)from elsewhere (EF size >BC N );stocks do not appear to be liquidated (EF depth =1).When a Footprint size is imported from outside without a corresponding use of Footprint depth,raw materials and/or products refined by means of renew-able resources are likely to be imported;it remains to be seen how such products are produced in the exporting nation.All points on the horizontal line from BC N represent long lasting or durable conditions,as long as exporting countries continue to sustainably support the importing nation.However,in an increas-ingly resource constrained world,dependency on trade and thus the necessity to compete internationally for natural resources (bio-capacity)increases the risk of geopolitical,economic and social instability (Moore et al.,2010).The faster a nation shrinks its eco-logical surplus or shifts from a surplus to deficit condition,the sooner that entity must make decisions about managing bioca-pacity demand,energy efficiency and related quality of life.In a resource constrained world,running an ecological deficit might in some cases become a risk for a nation’s economy as it takes finan-cial resources (purchasing power)to net import natural resources from elsewhere.It therefore becomes important for each nation to monitor and understand the size of its ecological deficit as in some cases it can provide an indication of the nation’s exposure to eco-logical risks (Moore et al.,2010).Footprint time series assessment can also enable key decision makers (strategists)to make informed decisions designed to avoid internal instability.Nations belonging to this case are,for example,Morocco and Jordan.Case C :for countries in this situation,Footprint size is lower than BC N (EF size <BC N ),even if a Footprint depth appears (EF depth >1).This means that local resources could be used more effectively without compromising local and/or global stocks.The presence of the Footprint depth term is synonymous with the use of highly refined and energy intensive materials (locally extracted and/or imported).By using stocks,any nation in this category not only appropriates resources from future generations but also con-tributes to the depletion of our finite natural capital thus affecting the Earth’s ability to provide for humanity in the long term.Most countries in this category have a high population density and/or a high level of industrialization but generally have a high per capita BC value even if not enough to compensate the Footprint value.Countries in this situation are rich and economically competitive countries,which choose to safeguard their local natural capital and rather use (and/or overuse)global stocks because of their import of energy intensive commodities and high emission of carbon dioxide;USA is the typical country belonging to this group.Case D :countries in this situation are characterized by an Eco-logical Footprint size higher than BC N and a Footprint depth higher than 1.This is due to both overconsumption of local stocks and import of both flows and stocks (e.g.carbon uptake land).From a resource perspective this is the most risky of the four cases.In this situation,local resources are exhausted and a big portion of EF is compensated by importing size and liquidating local and/or global stocks (adding depth).As such all three ways to compensate ecolog-ical deficit,as reported above,are used.As for case C,nations in this category not only appropriate resources from actual and future gen-erations but,by demanding stock of natural resources,they draw down the natural capital that allows our planet to sustain human life.This is a characteristic condition of industrialized and densely populated countries,where the lack of ecological space is a limiting。
工业过程生态生命周期评价的生态累积_耗模型
2.2
社会经济投入
社会劳动力的需求是工业过程不可忽略的因素 . Odum[10]和 Ukidwe 等[18]认为, 人力资源(HC)是人类 经济体特有的对于工业过程的投入 , 由此通过劳务 者的劳务工时(H)和技术水平参数来计量人力资源投 入的 ECECHC:
ECECHC Money P, k TrH H k aver Money P
工业过程生态生命周期评价的生态 累积㶲耗模型
钱宇*, 杨时颖, 杨思宇
华南理工大学化工学院, 广州 510640 *通讯作者, E-mail: ceyuqian@ 收稿日期: 2014-05-29; 接受日期: 2014-06-24; 网络版发表日期: 2014-08-15 doi: 10.1360/N032014-00157
这些成本所代表的社会资源共同维持了工业生 产所需要的直接或间接劳动力, 而其 ECEC 可以根据 生产成本和经济体中货币平均太阳㶲当量系数(EMR) 来确定. EMR 是社会经济体所消耗太阳㶲 当量总量 与货币流通量总量(以 GDP 计)的比值, 反映了一个 经济体的总体运行状况以及货币的实际购买力[10, 18]. Yang 等[23]曾对中国经济体的 EMR 进行过深入研究. 对此 , 本文提出以过程总的生产成本 (MoneyC) 来计 量工业过程社会经济投入的 ECECSC: ECECSC EMR Money C (4) 该计量方案全面考虑过程对象所需直接劳务和 购置资源的间接劳务对于太阳 㶲 当量的消耗 . 社会 经济投入的 ECECSC 计量可以利用过程的最终生产成 本数据, 数据较易采集, 不需再费力地追溯各个工业 上游过程.
2.1
自然资源消耗
在自然资源消耗(NC)方面, Hau 等[15]详细论述了 ECECNC 的计量方案, 主要分为可再生资源和不可再
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
1IntroductionUrban areas are increasingly important components of the landscape,with nearly 50%of the world's population and 75%of developed nations'population living in cities (United Nations,2002).Urban areas also have biodiversity value and contribute to natural processes such as population persistence and fire spread.However,it is unclear how the morphology of urban areas influences ecological processes,particularly the movement of organisms and the spread of environmental processes.One approach to measuring this influence is through the assessment of patterns of ecological connectivity important for these processes.In this paper I examine opportunities for using measures of urban form to predict ecological connectivity .The results contribute to the identification of urban areas likely to benefit from restoration and conservation activities and anticipate ecological prob-lems such as the spread of undesirable organisms or catastrophic processes such as fires exacerbated by particular urban forms.A metric is also developed to estimate ecological connectivity for other urban areas on the basis of aspatial information typically available for cities and regions.2Movement in urban landscapesMany researchers have addressed the issue of human mobility in urban environments (Hillier,2002;Srinivasan and Ferreira,2002;Waddell,2000).An equal challenge lies in understanding the mobility of organisms and the spread of environmental processes.Predicting ecological connectivity in urbanizing landscapesBritta G Bierwagen ôDonald Bren School of Environmental Science and Management,University of California,Santa Barbara,CA 93106,USAReceived 17May 2004;in revised form 3November 2004Environment and Planning B:Planning and Design 2005,volume 32,pages 763^776Abstract.Nearly half the world's population lives in urban centers,and these areas are increasingly important components of regional and global land cover.However,their ecological attributes are often overlooked,despite the presence of species,ecosystem services,and risks associated with the spread of pests or threatening processes such as fire.Movement and dispersal of organisms contribute to species persistence in urban landscapes;however,landscape patterns that promote ecological connectivity may also facilitate the spread of undesirable organisms or processes.I investigate how urban form can be used to predict ecological connectivity and assist in prioritizing urban landscapes for conservation activities and risk management.I examine the value of qualitative and quantitative descriptions of urban morphology as predictors of ecological connectivity by comparing sixty-six cities in the USA.Results show that qualitative categories are not adequate for describing ecological connectivity;multivariate descriptions are much better predictors,with urban area,number of urban patches,urban patch extent,level of aggregation,and perimeter area fractal dimension composing the significant synthetic variables.The dominance of area as a differentiating variable led to the development of a new urban connectivity index using a combination of urban area and state population size.This metric,based on readily available aspatial data,explains 78%of variation in ecological connectivity.These results provide a simple but novel tool for beginning to understand the role of urban morphology in promoting desirable environmental outcomes and managing environmental risks in urbanizing landscapes.DOI:10.1068/b31134ôPresent address:Global Change Research Program,National Center for Environmental Assessment,ORD ^US EPA (MC 8601N),1200Pennsylvania Avenue NW ,Washington DC 20460,USA,e-mail:bierwagen.britta@The encroachment of urban areas on natural habitats has many detrimental impacts,such as habitat loss,fragmentation (Alberts et al,1993;Bolger et al,1997;Brown et al,2000;Souleet al,1988;1992;Swenson and Franklin,2000;van Dyck and Matthysen,1999),edge habitat creation (Moran,1984),invasive species influx (Crooks and Soule,1999;Drayton and Primack,1996;Guilden et al,1990;Kowarik,1990;Rapoport,1993;Wetterer,1997;Zipperer et al,1997),and increased pollution (Charbonneau and Kondolf,1993).However,many species persist in urbanizing landscapes,sometimes utilizing new habitats and even expanding their ranges (Gompper,2002;Sol et al,1997).This means that the impacts of urbanization are not uniformly negative.In this paper I focus on the consequences of urbanization for ecological or environmental processes related to movement,and therefore connectivity (table 1).The increasingly common wildland ^urban interface (Platt,2001)contributes to the positive and negative effects of urbanization on eco-logical and environmental processes linked to habitat connectivity .This suggests a need to couple the understanding of urbanization with an assessment of the potential for organism movement and process spread.One approach is to examine the spatial structure of urbanization through morphology and its impact on ecological connectiv-ity,which is directly linked to species persistence through movement and dispersal and to the spread of other spatial processes such as fire.Movement is critical to organism persistence (Hanski,1994)because it includes foraging,mate location,territory defense,dispersal,and migration and is a product of the intrinsic qualities of organ-isms and their interaction with landscape features (Bierwagen,2003).The product of this interaction can be analyzed by means of connectivity metrics.The most widely used metrics examine the spatial distribution of habitat and provide estimates of howTable 1.Ecological and environmental processes affected by urbanization through changes in connectivity .Type of impactMechanismReferenceNegative effects of urbanizationSpread of invasive species Increased connectivity betweenurban and natural landscapes increases movement potential Aragon and Morales (2003),Boet et al (1999),Lim et al (2003),Woo and Zedler (2002)Spread of pest speciesIncreased connectivity between human-dominated and natural landscapes increases movement potentialNowak and McBride (1992)Spread of disease through population Fragmentation and decreased connectivity foster outbreaks in local populations Hess (1994),Lafferty and Gerber (2002),McCallumand Dobson (2002),O'Neill et al (1992)Fire spread and ignition Increased access and conducive spatial patterns in landscapes allow fire spreadPlatt (2001),Russell andMcBride (2003)Increased extinction risk Fragmentation decreases immigration and emigration from populations (metapopulation structure)Fahrig and Jonsen (1998),Hanski (1994),Hardy and Dennis (1999)Positive effects of urbanization Increased habitat,and range expansion Expanding urban areas and margins increase connectivityand habitat amount forgeneralists or urban adaptive speciesGompper (2002),Schumaker et al (2004),Sol et al (1997)764B G BierwagenPredicting ecological connectivity in urbanizing landscapes765successfully organisms with particular characteristics can move between habitat patches and across the landscape(Fahrig and Merriam,1985;King and With,2002).Relevant organism characteristics can include dispersal distance,dispersal rate,and movement behaviour(Moilanen and Nieminen,2002;Vos et al,2001).Urbanizaton can also disrupt the spread of physical processes such as fire across landscapes(Duncan and Schmalzer, 2004).Connectivity metrics can provide a proxy for the assessment of impacts of urbanization patterns on organism movement and other ecological processes,as evident from the examples in table1.3Urban morphologyThe footprint of urban areas has changed over time.Up to World War2cities were generally monocentric,with a central business district,surrounding housing,and transportation networks connecting cities(Makse et al,1995;Wu,1998).After World War2,patterns became more dispersed and decentralized(Aguilar,1999;Garreau, 1991;Makse et al,1995;Y eh and Li,2001).Researchers have periodically attempted to categorize these general forms(Medda et al,1998;Mesev et al,1995);Snellen et al (2002)qualitatively defined urban areas as concentric,lobed,linear or concentric polycentric,linear,or grid shaped(figure1,over).Several studies relate urban form to broader measures of environmental performance such as pollution,energy con-sumption,biodiversity,and human well-being(Alberti,1999;Calthorpe,1993)but generally do not relate landscape-level patterns to ecological processes at that scale.Quantitative approaches have been taken to attempt to differentiate urban areas from natural vegetation patterns and from one another by using various pattern analysis metrics commonly applied in landscape ecology research(Jenerette and Wu, 2001;Shen,2002).Shen(2002)examined the fractal dimension of twenty US cities by using a box-counting algorithm.Research had shown that urban areas have fractal properties,but the results had not been applied to relationships to other urban data. Shen(2002)showed that there is a strong correlation to overall urban areas.However, the measure of overall fractal dimension of an urban area did not distinguish different types of urban forms and is a more difficult metric to relate directly to ecological processes.Another method of defining urban form is the space syntax approach,which uses connectivity graph patterns to derive human behavior(Hillier,2002;Jiang et al, 2000).This approach uses free space,such as roads,as the basis for connectivity graphs,whereas previously mentioned methods use the built environment,especially through remote sensing,to derive urban forms.These,and other quantitative approaches have provided more detailed descriptions of differences between urban areas,but results have not been linked to impacts on specific ecological and envi-ronmental processes.In this paper I address this gap by comparing the utility of qualitative categories and quantitative metrics for describing aspects of urban mor-phology at a landscape scale relevant to the movement of organisms and the spread of environmental processes such as fire.An understanding of the effect of urban form on ecological and environmental processes is integral to the development of sustainable urban areas that contribute to biodiversity conservation,one of the goals of urban sustainability(Botequilha Leita o and Ahern,2002).4Selection of urban areasI selected a sample of sixty-six urban areas of different sizes and morphologies within 50kmÂ50km(2500km2)landscapes from across the USA from the National Land Cover Database(NLCD;Vogelmann et al,2001).Using ArcView GIS(ESRI,1999) I reclassified all NLCD developed landcover classes,including residential,commercial, industrial,and transportation,into a single urban class,to depict the overall formMonocentricSquare or linearLobedPolycentricLinear polyentricSprawlCategory Idealized form ExampleFigure 1.Idealized urban forms,adapted from Snellen et al (2002),used to classify urban areas qualitatively,including examples in each category.Urban areas in example landscapes are white;all other landcover is gray .766B G BierwagenPredicting ecological connectivity in urbanizing landscapes767of developed areas.I resampled the grids from30m to100m cell sizes by means of ArcGIS(ESRI,2002).This resulted in landscapes with two landcover classes:urban and habitat.The habitat class includes natural landcover as well as parks or other open space.I selected urban areas that were not dominated in form by natural features such as large lakes,oceans,or mountains,which heavily constrain the development pattern (but see Shen,2002).River features were largely unavoidable but tended not to dom-inate the overall shape.I selected urban areas in different parts of the USA based on differences in road network patterns,as these can also influence the overall urban form (Snellen et al,2002).For example,in the Midwest,roads generally follow a grid pattern,whereas in the Northeast roads are denser and more complex.Cities varied in spatial extent,from42X06km2(Helena,MT)to1736X47km2(Los Angeles,CA),and in human population size,from23564(San Fernando,CA)to3694820(Los Angeles; see table2,over).Population statistics are for the cities on which the landscape selection was centered and generally underestimate the total population size within the developed areas of the2500km2landscapes.However,state population was also used,which may be more appropriate for the regional landscapes used in this study.5Landscape analysisI qualitatively classified the sixty-six urban areas into the forms shown in figure1, using slightly modified categories from those in Snellen et al(2002;table1).I defined sprawl as an urban area that exceeds the extent of the landscape size used and therefore has no discernable form.I combined linear with concentric cities and combined both polycentric categories because of the small sample sizes and visual similarity.Although I did this classification before the quantitative analysis,I also asked six other scientists not involved with this research to apply the categories to the sixty-six urban areas to reduce classification bias.I then selected the urban areas that were all categorized in the same way or that differed by only one.In the analysis,I used both the classification for all sixty-six urban areas and the consensus classification.Each urban landscape was analyzed with a suite of landscape metrics selected on their potential usefulness in distinguishing different types of urban forms and patterns (Goldstein et al,2004;Jenerette and Wu,2001;Luck and Wu,2002),their predictable response in landscapes,and the nonredundancy of the metrics(Neel et al,2004).I used Fragstats software(McGarigal and Marks,1995)to calculate all metrics for the urban areas.I measured urban area(CA)in hectares;the number of urban patches(NP),as defined by an eight-cell neighborhood by cover class;perimeter-area fractal dimension (PAFRAC)for shape complexity,with values between1and2,where simple shapes approach1,and more complex and plane-filling perimeters approach2,calculated by regressing the logarithm of patch area against the logarithm of patch perimeter and dividing2by the resulting slope;class aggregation(CLUMPY),which calculates the proportional deviation of the cell adjacencies with the same cover class from a random spatial distribution,whereÀ1is maximally disaggregated,0is random,and approx-imately1when the cover class is maximally aggregated;and the area-weighted mean radius of gyration(GYRATE.AM)for urban patch extent,also defined by an eight-cell neighbor rule,to measure the distance between each cell in the patch and the center of the patch.Area-weighted means generally correlate better with dispersal success (Schumaker,1996).I used the Fragstats CONNECT function,[see equation(1)below],to calculate the ecological connectivity of the habitat remaining in these urban landscapes.This resulted in an estimate of the potential for successful interpatch movement given a dispersal of500m.768B G BierwagenTable2.Selected urban areas,including connectivity index(C H h),measured habitat connectivity (C h),urban form classification(see figure1),inclusion as a consensus city(`yes'or`no'),corre-sponding cluster category,total urban area,and local population.The table is sorted from highest to lowest C H h.C H h is calculated by means of equation(2)(see text,section9)and any negative values are defined as zero.City and Abbrevia-Abbrevia-C H h C h Form Con-Cate-Area Popu-state tion in tion in sensus gory(km2)a lation b figure2figure3Helena,MT mt_helen mthl 3.36 3.82concentric no142.0625780 Minot,ND nd_minot ndmn 3.16 4.30polycentric yes257.2236567 Casper,WY wy_caspe wycs 3.12 4.32lobed yes263.8149644 Bismark,ND nd_bisma ndbs 3.03 3.23polycentric no465.9455532 Rapid City,SD sd_rpdct sdrp 2.89 3.69lobed no472.9359607 Cheyenne,WY wy_cheyn wycy 2.82 4.00concentric yes288.2853011 Sioux Falls,SD sd_suxfl sdsx 2.56 2.15concentric yes2104.08123975 Mason City,IA ia_masnc iams 2.54 4.32polycentric yes374.0629172 Pine Bluff,AR ar_pineb arpn 2.47 2.03concentric yes281.5655085 Bloomington,IL il_blmng ilbl 2.35 2.94concentric no262.2764808 St Cloud,MN mn_stclo mnst 2.20 2.27lobed no493.1159107 Decatur,IL il_decat ilde 2.17 2.88lobed no474.7381860 Iowa City,IA ia_iacit iaia 2.16 2.71polycentric no3111.2462220 Urbana,IL il_urban ilur 2.14 2.22concentric no477.0136395 Topeka,KS ks_topek ksto 2.14 1.75concentric yes4116.12122377 Waterloo,IA ia_wtrlo iawt 1.95 1.51concentric no4138.568747 Eugene,OR or_eugen oreu 1.91 1.11lobed no2139.07137893 Springfield,IL il_sprgf ilsp 1.87 1.73concentric no4102.89111454 Lexington,KY ky_lxngt kylx 1.84 1.28polycentric yes4143.52260512 Salem,OR or_salem orsa 1.83 1.15polycentric yes4151.22136924 Cedar Rapids,IA ia_cdrra iacd 1.78 1.49concentric no4165.66120758 Madison,WI wi_madis wima 1.74 1.15polycentric yes2148.15208054 Muncie,IN in_munci inmn 1.61 1.07polycentric yes5164.0667430 Springfield,MO mo_sprgf mosp 1.60 1.06concentric no4168.83151580 Fort Wayne,IN in_ftway inft 1.460.84lobed yes4192.63205727 Portage,MI mi_porta mipo 1.45 1.22polycentric yes4171.5144897 Bakersfield,CA ca_bakrf caba 1.39 1.51concentric no4131.77247057 Rockford,IL il_rockf ilrk 1.370.90concentric no4175.64150115 Peoria,IL il_peori ilpe 1.36 1.10lobed no5176.33112936 Lansing,MI mi_lansi miln 1.350.88lobed no4189.28119128 Colorado co_cospr coco 1.310.66concentric yes4247.76360890 Springs,COWichita,KS ks_wichi kswi 1.300.55concentric no4282.83344284 Des Moines,IA ia_dmnes iadm 1.270.59concentric no4285.64198682 Little Rock,AR ar_ltroc arlt 1.130.42polycentric no5338.26183133 Spokane,WA wa_spokn wasp 1.100.64lobed yes4282.9195629 Grand Rapids,mi_grdrp migr 1.040.68lobed yes4262.86197800 MILas Vegas,NV nv_vegas nvvg 1.000.32concentric no4419.16478434 Modesto,CA ca_modes camo0.970.81polycentric yes5206.61188856 Salt Lake City,ut_saltl utsl0.960.33lobed yes5424.57181743 UTFlint,MI mi_flint mifl0.910.62lobed no5303.98124943 Stockton,CA ca_stokt cast0.890.71polycentric yes5223.35243771 Tucson,AZ az_tucso aztc0.870.44concentric yes4376.55486699 Worcester,MA ma_wrcst mawr0.870.70lobed no5356.72172648a Area classified as urban within2500km2landscape.b Source:US Census2000;see .I chose this distance based on the mean dispersal distances for several butterfly species,other specialist organisms,and firebrand spread (Bierwagen,2003).CONNECT is defined as:C i 100Â12n i n i À1 ÃÀ1nj b kc i j k ,(1)where c i j k is the `joining'between patch j and k ,with 0corresponding to unjoined and 1to joined,of the corresponding patch type,i ,based on a defined dispersal distance in meters,and n i is the number of patches in the landscape of the corresponding patch type (McGarigal et al,2002).In the case of habitat,i h,and therefore C h provides an estimate of the fraction of habitat patches that can be reached in one movement step.It can be interpreted as the fraction of patches available to a dispersing organism or the fraction of patches susceptible to a spreading pest or fire in the landscape.I also calculated the connectivity of urban patches,C u ,where i u,in each landscape.6Analysis of connectivity response to urban formI used a suite of simple and multivariate statistics to examine the relationship between ecological connectivity and the qualitative urban form categories and the quantitative landscape metrics.The goal was to find the elements of urban morphology that best predict ecological connectivity .I analyzed the consensus city categories by usingTable 2(continued).City and Abbrevia-Abbrevia-C H h C hFormCon-Cate-Area Popu-statetion in tion in sensus gory (km 2)a lation bfigure 2figure 3Tulsa,OKok_tulsa oktl 0.700.51lobedno 5498.26393049Fresno,CA ca_fresn cafr 0.690.54concentric no 4277.5527652Oklahoma City,ok_okcit okok 0.680.68lobed no 5510.6506132OKDayton,OH oh_daytn ohdy 0.490.32lobedyes 5454.89166179Austin,TX tx_austi txau 0.410.61concentric no 5422.27656562Hartford,CT ct_hrtfr ctht 0.410.26polycentric no 5681.38121578Columbus,OH oh_clmbu ohcl 0.390.30concentric yes 5509.31711470Indianapolis,IN in_inpol inin 0.340.25lobed no 5633.91781870Pittsburgh,PA pa_pitts papt 0.250.49lobedno 5578.79334563Batavia,IL il_batav ilba 0.240.26polycentric no 5580.8423866Denver,CO co_denv codn 0.070.14concentric yes 5921.32554636San Fernando,ca_sanfecasf0.000.22concentricno 5581.5623564CAOrlando,FL fl_orlnd flor 0.000.19concentric no 5814.9185951Minneapolis,mn_mnapl mnmn 0.000.13concentric no 51116.37382618MNSacramento,CA ca_sacmn casc 0.000.20concentric yes 5630.54407018San Antonio,tx_snant txsa 0.000.30concentric yes 56701144646TXPhoenix,AZ az_phoen azph 0.000.17concentric yes 51005.341321045Atlanta,GA ga_atlnt gaat 0.000.25sprawl yes 5945.18416474Fort Worth,TX tx_ftwor txft 0.000.21polycentric no 5824.56534694Riverside,CA ca_rivsi carv 0.000.10sprawl yes 51161.91255166Dallas,TX tx_dalla txdl 0.000.14sprawl no 51233.571188580Los Angeles,ca_lahcala 0.000.06sprawlyes 51736.473694820CAHouston,TXtxhu0.000.08sprawlyesna1288.51953631Predicting ecological connectivity in urbanizing landscapes 769ANOVA to predict ecological connectivity to see if similar visual forms have similar connectivity.Then I used the qualitative urban form classifications as a priori catego-ries for a multiresponse permutation procedure (MRPP;McCune and Mefford,1999).The MRPP calculates an average distance between the assigned groups,in this case the qualitative form categories,and their response,the quantitative metrics measured to differentiate urban areas.If the response variables cluster by the specified groups,then the average intragroup distance will be small compared with the average distances from other possible combinations.I also examined the qualitative categories in a cluster analysis by using relative Euclidean distance as the distance measure and the farthest neighbor as the group linkage (McCune and Mefford,1999).The cluster analysis gives a visual representation of which cities are more similar to one another,based on the quantitative data.If the quantitative information gives the same result as the qualitative categories,the clusters and categories would correspond to one another.The measure of clustering is based on quantitative metrics.The quantitative landscape metrics allowed an assessment of which aspects of urban morphology could be used to distinguish the urban areas from one another.This was done by collapsing the six urban form metrics by means of a principal components analysis (PCA)in a correlation matrix.This provided a synthetic measure of urban morphology that could be correlated with ecological connectivity .I used the principal component scores as predictors of ecological connectivity (C h )in linear regression models.7ResultsMany urban areas were classified as concentric forms (twenty-eight overall,including eleven consensus cities;see table 2).The MRPP shows that the qualitative classifications are more distinct from one another than expected by chance,for all sixty-six urban areas (p `0X 001)and for the consensus cities (p `0X 001).However,intragroup heterogeneity (A 0X 16)approaches the amount expected by chance (A 0)for all sixty-six urban areas,suggesting that the qualitative classification does not capture the full variation in C h .The consensus cities have less intragroup heterogeneity (A 0X 42),but the qual-itative categories still do not encompass all of the variation.Further investigation with the use of cluster analysis showed that the a priori groups do not cluster together,either for sixty-five urban areas (figure 2)or for the thirty consensus cities.Both analyses produced five groups that cluster predominantly by urban area size (table 2).PCA shows that a combination of CA ,GYRATE.AM ,and NP organize the cities along the first axis and explain 40.8%of the variance between urban areas [figure 3(b),over].The second component axis explains a further 25.9%of the variance.This component is composed of approximately half CLUMPY (positive)and half PAFRAC (negative).Axis three explains another 19.4%,with PAFRAC and C u (connectivity of urban patches)as the main components.The axis scores show that total urban size,urban patch extent,and number of patches are the strongest measures that differentiate these cities from one another [figures 3(a)and 3(c)].Other quantitative measures of urban form are less important in distinguishing one city from another.In a linear regression,the first two principal component scores predict 50%of the variation in C h between urban areas (R 2 0X 50;P `0X 001;F 263 31X 94 .This result shows that some aspects of urban form measured quantitatively can predict a significant proportion of the remaining ecological connectivity across urban areas.8Qualitative versus quantitative description of urban formsAlthough the analysis of the qualitative categories shows that there is some similarity between urban areas classified visually into these four categories,they do not inform us about ecological or environmental connectivity.These categories are not useful in,770B G Bierwagenpredicting the ease or difficulty of organism movement or the risks of the spread of catastrophic processes such as fires through urbanizing landscapes.Therefore,simple form categories are less useful for planning sustainable development than are easily calculated quantitative measures.Quantitative,multivariate,descriptions of urban areas correlate more closely with ecologically relevant connectivity metrics.These factors should be considered in urban and regional planning to increase compatibility with ecological processes and sustainable development goals.PCA results suggest that it is possible for urban areas that appear visually distinct to have similar ecological connectivity scores.For the cities considered in this study,urban area,extent,and the number of urban patches are the most important factors in determining ecological connectivity .These characteristics of urban areas may also relate to other ecological impacts such as declines in edge-sensitive species (Bender et al,1998),declines in species requiring specific disturbance regimes (for example,fires,floods;see Brawn et al,2001),or increases in invasive species (Aragon and Morales,2003;Boet et al,1999;Lim et al,2003;Woo and Zedler,2002).9Other urban descriptors and ecological connectivityAlthough the quantitative descriptions of urban morphology shows a correlation with ecological connectivity,this relationship is not necessarily more useful than if one were to calculate ecological connectivity directly,as the urban metrics are derived by using the same landscape analysis program.The qualitative form categories would hypothet-ically provide a simple way to estimate ecological connectivity.However,the formsandDistance (objective function)Information remaining (%)2Â10À66X 4Â10À21X 3Â10À11X 9Â10À12X 6Â10À1Figure 2.Dendrogram showing the clustering of cities (for key see table 2),based on landscape metrics.The symbols designate the a priori classification of urban areas into forms (see figure 1),and the groups are those obtained from cluster analysis.Predicting ecological connectivity in urbanizing landscapes 7710.20.0À0.2C o m p o n e n t 2642À2À4À6(a)(b)(c)ComponentComponent 3Figure 3.Principal components analysis of class-level metrics of urban areas:(a)biplot of component axes 1and 2,where the arrows indicate the relative strength of the axis component;(b)barplot of the proportion of variance explained by each component axis;(c)barplot of significant variables in first three component axes.Note:for definitions of CA ,CLUMPY ,CONNECT ,GYRATE.AM ,NP ,and PAFRAC ,see text,section 5.For key to cities see table 2.772B G Bierwagen。