Environ Monit Assess (2009) 157:137=149 DOI 10.1007/S10661-008-0523-2 Evaluating the condition of a mangrove forest of the Mexican Pacific based on an estimated leaf area index mapping approach J.M, Kovacs - J. M. L, King « F. Flores de Santiago « F. Flores-Verdugo Received: [1 December 2007 / Accepted: 11 September 2008 / Published online: 21 November 2008 (© Springer Science + Business Media BV. 2008 Abstract Given the alarming global rates of man- grove forest loss it is important that resource managers have access to updated information re- garding both the extent and condition of their mangrove forests. Mexican mangroves in par= ticular have been identified as experiencing an exceptional high annual rate of loss. However, conflicting studies, using remote sensing tech- niques, of the current state of many of these forests may be hindering all efforts to conserve and manage what remains, Focusing on one such system, the Teacapdn—Agua Brava-Las Hacien- das estuarine-mangrove complex of the Mexican Pacific, an attempt was made to develop a rapid method of mapping the current condition of the mangroves based on estimated LAL Specifically, using an AccuPAR LP-80 Ceptometer, 300 in- direct in situ LAT measurements were taken at J.-M. Kovacs (3) J. M.L King Department of Geography, Nipissing University, ‘North Bay, ON, PIB 8L7, Canada ‘e-mail: Johnmk@nipissingu.ca F. Flores de Santiago Facultad de Ciencias Marinas, Universidad Autonoma. de Baja Califomia, Ensenada, Baja Califomia, México F. Flores-Verdugo Instituto del Giencias del Mar y Limnolegfa, Universidad Nacional Autnoma de Mexico, ‘$2000 Mazatiin, Sinaloa, México various sites within the black mangrove (Avicen- nia germinans) dominated forests of the north= em section of this system. From this sample, 225 measurements were then used to develop linear regression models based on their relationship with, corresponding values derived from QuickBird very high resolution optical satellite data. Specif- ically, regression analyses of the in situ LAI with both the normalized difference vegetation index (NDVI) and the simple ration (SR) vegetation index revealed significant positive relationships [LAI versus NDVI (R? = 0.63); LAT versus SR (R = 0.68)]. Moreover, using the remaining sam- ple, further examination of standard errors and of an F test of the residual variances indicated little difference between the two models. Based on the NDVI model, a map of estimated mangrove LAI was then created, Excluding the dead mangrove areas (ise, LAI = 0), which represented 40% of the total 30.4 km? of mangrove area identified in the scene, a mean estimated LAI value of 2.71 was recorded. By grouping the healthy fringe mangrove with the healthy riverine mangrove and. by grouping the dwarf mangrove together with the poor condition mangrove, mean estimated LAI values of 4,66 and 2.39 were calculated, re~ spectively. Given that the former healthy group only represents 8% of the total mangrove area examined, it is concluded that this mangrove sys tem, considered one of the most important of the Pacific coast of the Americas, is currently D springer138 Environ Monit Assess (2009) 157:1374149 experiencing a considerable state of degradation, Furthermore, based on the results of this investi~ gation it issuggested that this approach could pro vide resource managers and scientists alike with avery rapid and effective method for monitoring, the state of remaining mangrove forests of the Mexican Pacific and, possibly, other areas of the tropics. Keywords Mangrove condition LAI + Ceptometer + QuickBird + Mapping + Monitoring « Mexico Introduction ‘The effort to develop cost-effective techniques for rapid monitoring of mangrove forests has been hastened by updated reports of global mangrove forest loss (Valiela et al. 2001). Not surprisingly, in an attempt to curtail these alarming rates of deforestation many resource managers are invest ing in remotely sensed data from satellites as a ‘means of mapping the extent of remaining man- grove forests (Brondizio et al. 1996; Gao 1998; Green et al. 1998; Webb et al. 2000; Tong et al 2004), When historical scenes are available these data are also being used to examine the causes of these losses (Kovacs et al, 2001; Alonso- Perez et al, 2003; Hemdndez-Comejo et al. 2005; Muttitanon and Tripathi 2005; Beland et al. 2006; ‘Thu and Populus 2007). Given their continuous records, extensive coverage and cheaper costs in comparison to more recent very high resolution Earth observing satellites (eg. QuickBird and IKONOS), the tradition sensors, chiefly from the Landsat and SPOT series, have remained the pop ular selection for such endeavors. However, given their limited spatial resolution, these investiga tions have, with some exception, been limited to basic qualitative mapping of these forests (see review in Green et al. 1998). Moreover, a vast majority of these studies have focused simply on mapping these forests at a very generic, aggre gated level. As a result it is often too difficult to determine whether the condition of a mangrove forest is actually improving over time. For ex- ample, using a single mangrove class, several re~ searchers, using Landsat data, have reported that, D springer the mangroves of the Teacapsin region located within the state of Sinaloa, Mexico, have recently expanded (Berlanga-Robles and Ruiz-Luna 2002; Fuente and Carrera 2005). Conversely, Kovacs et al, (2001), also using Landsat data but employ ing a more refined classification scheme based on the condition of the mangroves, have reported that the system is undergoing extensive degra dation, As a result of such limitations, many have suggested that mapping quantitative data (ic. biophysical parameters) of these forests from optical satellite data may be more useful and appropriate when managing these ecosystems. In patticular, optical satellite data has become ex- ‘tremely useful in monitoring leaf area index (LAI) for a plethora of vegetation types worldwide (Asner et al. 2003). In recent years, this para meter has become a principal descriptor of veg- etation condition for a variety of physiological, climatological and biogeochemical studies (Asner et al. 1998). Although extremely limited in num- ber, studies have indicated that mangroves LAI, or canopy coverage, can be successfully extracted using the traditional sensors (Jensen et al. 1991; Ramsey and Jensen 1996; Green et al. 1997). However, given the coarse spatial resolution of these sensors in comparison to the more recent series of very high spatial resolution satellites, (eg. IKONOS, QuickBird), it is suggested that these later data could be more valuable in map- ping mangrove LAL. In fact, given the increased spatial resolution of these new sensors, mangroves, have, for the first time, been mapped accurately at the species level using satellite data (Wang et al. 2004a, b; Dahdouh-Guebas et al. 2005; Kovacs et al. 2005). With regards to mapping mangrove LAT from very high resolution satellite data, Kovacs et al. (2004) developed statistical models based on the relationship of in situ LAI collected using an LAI- 2000 PCA (LI-COR, Lincoln, Nebr.) and vegeta- tion indices derived from IKONOS data for the Agua Brava Lagoon of the Mexican Pacific, One of these models was then used to map the LAI of the entire scene and subsequently separate the ‘two species of mangrove found in this study area (Kovacs et al. 2005). Although the LAT-2000 PCA, was deemed an appropriate tool for such endeay= ors and continues to be used as a standard toolEnviron Monit Assess (2009) 157:137=149 139 for indirect measures of vegetation LAI, Kovacs ically, the LAI-2000 PCA is dependent on dif- et al, (2005) did indicate some limitations with fuse radiation for the most accurate measure~ its application for mangrove monitoring. Specif- ments, Consequently, if homogeneous cloud cover Fig 1 Location of the sity aca within the vague guy ar \ > Bite Briva-Las Haciendss < cestuarinosmangrove », system, Mexican Pacific ( \ \ steno \ mer ar -\\ \ XY PACIFIC OCEAN aes D springer10 Environ Monit Assess (2009) 157:1374149 is not present then it is recommended that the instrument be used in dusk or dawn conditions when direct solar radiation is absent. Although these periods can be quite lengthy in temperate areas, in the tropics they may be extremely brief, For example, Kovacs et al. (2005) indicated that for their study area the measurements could only be preformed one hour before sunrise or one hour after sunset. Given the logistical difficulties associated with collecting any biophysical mea- surements of mangrove forests, it is hoped that, an alternative instrument could be employed in order to provide scientists with a more flexible and. possibly more cost-effective indirect measure of in situ mangrove LAL The purpose of this investi- gation is to examine whether an alternative indi- rect optical radiometric tool, the AccuPAR LP- 80 Ceptometer (Decagon Devices, Inc., Pullman, Wash.), could also be used in association with very high resolution satellite data to map man- grove forest LAI of the Teacapan region, Mexican, Pacific, Previous studies conducted in terrestrial forests (Pierce and Running 1988 White et al 1997; Hyer and Goetz 2004), have reported con- siderable success in using a Decagon Ceptometer for estimating LAI with some having compared it to the LAI-2000 PCA. However, unlike the LAI- 2000 PCA, the AccuPAR LP-80 works best in direct solar radiation, Consequently, in relatively persistent cloud free regions or during the dry season, this larger window of opportunity for data collection could be ideal especially when work- ing in the remote and often difficult conditions associated with mangrove forests. Since the focus of this study are the mangroves of the Teacapan region, the mapping of the estimated LAI would, moreover, resolve the conflicting accounts regard- ing the state of this extremely important forest. For Mexico, the loss of total mangrove area has been reported (Valiela et al. 2001) at approxi- mately 20% over a 9-year period and thus rapid approaches for effectively monitoring the condi tion these forested wetlands are urgently required, Site description ‘The focus of this investigation is the Teacapan— Agua Brava-Las Haciendas estuarine-mangrove D springer system of the Mexican Pacific which is still some mes refereed to as either the Marismas Na- cionales or the Teacapén—Agua Brava lagoon— estuarine system. Extending along the coastal states of Sinaloa in the north and Nayarit in the south, this mangrove forest is considered one of the largest of its kind on the western coast of the Americas (Fig. 1). The system supports three dominant species of mangrove: red mangrove (Rhizophoramangle), black mangrove (Avicennia germinans), and white mangrove (Laguncularia racemosa). The black and white dominate prin- cipally in the northem and southem sections of this system, respectively, Based on the Lugo and Snedaker (1974) classification scheme, large red fringe mangrove can be found throughout the sys tem along the edges of the numerous tidal chan- nels and lagoons, However, this species also forms extensive homogeneous stands in the southem section adjacent to the large white basin mangrove areas. In the north, neighboring a very narrow fringe (~10 m) of red mangrove, are relatively narrow zones of riverine mangrove (~75m) which are dominated by black mangrove, Further inland from these riverine mangroves are now extensive areas of dwarf or poor condition black mangrove, Although an important local renewable resource (Kovacs 1999), the mangroves of this system have experienced considerable degradation, Supported by the views of local fishermen (Kovacs 2000) it has been hypothesized by scientists (Flores Verdugo 1986; Flores-Verdugo et al. 1997) that hydrologic changes, principally changes in salin- ity levels and tidal regime, have adversely im- pacted these forests following the opening of a canal (Cuautla channel) in 1972. In an attempt, to mitigate such changes, the town of Francisco Villa, of the Agua Brava Lagoon, is currently con- structing a series of tidal channels to encourage restoration in dead mangrove areas that prolif- erated soon after the opening of the canal, Al- though unable to map at the species level, Kovacs et al, (2001), using multi-temporal Landsat im- agery, did record extensive areas of degradation in both the northem and southem sections of the ‘Teacapén-Agua Brava-Las Haciendas estuarine mangrove system, The extent of damage in the south, specifically the Agua Brava Lagoon, ‘was recently confirmed using mapped estimatedEnviron Monit Assess (2009) 157:137=149 Mi LAI values derived from much higher spatial resolution IKONOS data (Kovacs et al, 2005), Others (Berlanga-Robles and Ruiz-Luna 2002; Fuente and Carrera 2005) using Landsat data, have reported that the extent of mangrove cover age in the north (State of Sinaloa) has recently in- creased but, unlike Kovacs et al. (2001), they have not provided any indication as to the condition (ic-health) of these mangroves. However, Kovacs, et al. (2001) did not provide any true biophysical measures of the state of these mangroves, Con- sequently, the focus of this investigation is this northem section of this system, specifically the mangroves of Isla La Palma and of the adjoining, mainland (Fig. 1). Pool et al. (1977) had indicated in their investigation of this island in 1973 that it had, in comparison to other riverine mangroves, they encountered in the Americas, some of the tallest and most productive trees, with a reported basal area of 608 m2/ha, Data and methodology In situ LAT collection To estimate in situ LAT an AccuPAR LP-80 Ceptometer was employed. This device uses a O8 m long linear probe containing 80 sensors to measure the incoming photosynthetically active radiation (PAR) found between 400 and 700 nm. wavelengths, At least one reading must be taken above and below the canopy in order to esti= mate LAL The specific inversion equation used by this particular instrument to determine LAL is as follows; _ [= sk) i - tine LAl= Ta 0aThy a) ‘The PAR readings taken above and below the canopy are used to calculate tau (r). Specifically, tau is the ratio of PAR measured below the canopy to PAR above the canopy. The extinction coeflicient (K) is determined automatically by the LP-80 by using the latitude, longitude and the time of day to calculate the appropriate zenith angle ©). K= @ ‘This calculation of the extinction coefficient as- sumes the leaf angle distribution to be spherical. However, the user can, prior to data collection, alter the leaf distribution parameter from this de- fault setting, The beam fraction (fi,) component of the equation is simply the ratio between direct, radiation coming from the sun and radiation com- ing from all other sources and it to is calculated automatically by the LP-80. Finally, the constant, Aisdetermined by the instrument based upon the leaf distribution and the canopy leaf absorption qualities, For this device, this later component is, assumed to be 0.9 in all LAT sampling routines, ‘The AccuPAR LP-80 is equipped with an ex- temal PAR sensor which allows for simultane ous above and below canopy PAR readings and thus can simplify the LAT collection. However, the extemal cable provided with this sensor only measures 2 m in length and is consequently lim- ited for use in short canopies. An extension cable for the external sensor, measuring 7.6 m, can be purchased with the unit. However, for accurate readings it is recommended that the sensor be properly leveled, ideally on a tripod, and thus such an arrangement may not be feasible for rapid LAL collections in mangrove forests. Conse quently, and as suggested for such environments, (AccuPAR model LP-80 Operator’s. Manual, 2004), a separate datalogger with an attached PAR sensor was employed to collect the above measures. Specifically, a HOBO Micro Station was placed on a leveled tripod in a clearing near to the AccuPAR readings and the above canopy PAR measurements were taken every 20's. Once downloaded to a PC, the records between the two sensors were matched by time and the LAT calcu lation post-processed in a simple spreadsheet. To ensure utmost accuracy, lll readings were taken in direct sunlight during peak solar radiation hours, For this study area this meant that all collections, occurred between 1000 and 1400 hours. At each location, ceptometer measurements were taken in a circular fashion from a central position at 45 degree intervals for a total of eight readings per site. These eight measurements were then averaged, In total 300 LAI sites were recorded (ie. 2400 ceptometer readings) in the study area over a period of 9 days during the month of January, 2007. Based on a classified map derived D springer12 Environ Monit Assess (2009) 157:1374149 from Landsat data (Kovacs et al. 2001), there was an attempt to capture the in situ LAT from a range of mangrove conditions present in the system. Of the 300 sites analyzed, 97, 104 and 99 averaged LAI values were measured from what was considered to be poor condition, dwarf and. tall black mangrove canopies, respectively. These tall healthy mangroves were found only in the fringe and riverine mangrove areas limited to the edges of the estuaries and tidal creeks. In order to later match these LAT values with the satellite data, the central location of each data collection point was recorded using a Trimble GeoXT GPS hand held unit, This raw GPS data was then post, processed with Trimble Pathfinder software using, data obtained froma stationary Geo XT GPS unit (ie. base station), The average accuracy on these data was determined to be at a sub-meter level. Satellite imagery, data processing and statistical modelling QuickBird imagery of the study area was acquired for April 25, 2007 as close as possible to the dates, of field data collection and still within the dry season, Although both panchromatic and multi spectral images were acquired, at 60 x 60 om and 2.4 x 24 m pixel resolutions, respectively, only the multi-spectral data were used for this study, ‘The upper left comer of the image was located at 22°38'52" N and 105°42/17” W and the multi- spectral image measured 4575 rows by 2461 columns, To geometrically correct the image, six ground control points (GCPS) were collected dur= ing the field excursions. Specifically, edges of fish ing weirs and corners of fishing huts were selected. ‘The high spatial accuracy maintained though the use of the Trimble units provided ideal points for the geometric correction of QuickBird imagery. The QuickBird imagery was re-sampled using a nearest neighbor algorithm at a residual root, mean square error of 0.20 m (x-axis) and 0.18 m (axis). Following the geometric correction, a normal- ized difference vegetation index (NDVD and a simple ratio (SR) data layer were derived from the QuickBird multi-spectral imagery. Using the GPS data, pixel values from both vegetation indices were then extracted for all of the in situ LAI D springer sites, As a result, corresponding values of NDVI, SR, and AccuPAR LP-80 LAI could be matched for each location, It must be noted that some error most likely resulted from the collection sites, located in the southem section of the scene where avery fine cloud cover was present. From the 300, records, 225 were randomly selected to develop linear regression models of the vegetation indices by the in situ data, The remaining samples were used to test, using standard errors of estimate, the predicted results. A classification of the original four bands of multi-spectral QuickBird data was also preformed using segmentation procedures to create several masks in order to later extract mean. estimated LAL values for the specific condition of mangroves present as well as to calculate their extents, Specifically, masks were created to cover healthy condition mangroves, to cover dead man- grove areas and to cover combined dwarf and poor condition mangroves. The dead stand areas were confirmed using 1970 aerial photographs. Using the NDVI regression, all pixels under the mask were then converted to estimated LAI val ues. Any negative values were reverted to zero LA\ values and measurements of the mapped LAI were then derived from these data, Results ‘The in situ estimated averaged LAI values col- lected with the AccuPAR LP-80 ranged from O.11 to 7.53. A mean LAI value of 257 was recorded for these data. Although a few red and white mangroves were encountered in some of the healthy mangrove stands, the black mangrove was always dominant. Moreover, only the black mangrove was encountered in the poor condition and dwarf mangrove areas, Similar values of in situ LAI were recorded for these later two classes. However, due to their height, recordings using the LP-80 of the dwarf mangroves had to be taken closer to the ground. These individual mangroves, provided a more dispersed (i.e. clumped) canopy in comparison to the others even though closed canopies were selected. Locations of dead man- grove stands were recorded for ground truthing purposes but no LP-80 measurements were taken. in these stands,Environ Monit Assess (2009) 157:137=149 143 With regards to the LAI versus the vegetation indices, the two linear regression models devel oped all fit the data quite well, with fairly strong, observed R? values (**P < 0,001; 2 = 225) of 0.63 and 0.68 for the NDVI and the SR data, respectively (Fig. 2). The F-test for both models and ¢ test for the slope estimates indicate that the two linear models and the b coefficients are statistically significant, Based on the 75 randomly selected cases, the plotted results of the measured. in situ LAT versus the predicted LAI, based on the satellite data, also suggest both models as valid (Fig. 3). Standard error of estimates of 081 and 0.73 were recorded for the predicted LAI versus the actual LAI for the NDVI and the SR data, respectively. Further, to determine statistically whether one regression model was better for predicting LAI than the other, an F test, based on the variances of the residuals determined from this test sample, was conducted. Specifically, this test was based on. the ratio of the two observed sample variances Fig.2_ Scatter plots and ' fitted linear regression rt (ay lines of LAI from in situ measurements using an . y= 1276x232 AauPAR LPS) R=063 Ceptometer against the L normalized difference Aa ‘vegetation index (a) and 1 thesimple ratio (b) derived from QuickBird 2 satellite data for 225 sites y= 221-262 which were derived from the sum of the squared differences (or SSEs) between the observed and predicted values for each of the linear regression models. Given that, at a 95% confidence level, the Fest [73, 73, P = 0.05/2] value was 1.7 and the Figg (SSEs NDVU/SSEs SR = 90.1/69.3) value was 1.3, it is not possible to suggest any statistically significant difference between these two models (ies Foss < Fests Although either model could be employed, the NDVI regression was applied to the QuickBird data to map the estimated mangrove LAL for the entire scene (Fig, 4). The NDVI is less affected by variations in atmospheric condition (Mather 1987) and, moreover, in comparison to SR, the dynamic range isstretched in favor of low-biomass conditions (lensen 2007). Consequently, since this, system has experienced substantial degradation, the NDVI could be considered more applicable. Based on the mapped estimated LAI and exclud- ing dead mangrove areas (ie. LAI = 0), which represented 12.1 km? of the total 304 km? of NOVI D springer4 Environ Monit Assess (2009) 157:1374149 Fig.3 QuickBird ® predicted LAI based cn. the NDVI (a) and the simple ratio (b) versus in situ measured LAT using an AccuPAR LP-80 Ceptometer based on 75 E validation sites . i m a t e a L ; E t i m a t e a L A 1 mangrove identified in the scene, a mean man- groves, respectively. The healthy mangrove area grove LAI value of 2.71 was calculated. When mapped was quite small, only measuring 2.56 km?, further subdivided, LAT values of 4.66 and 2.39 in contrast to the combined dwarl/poor conditions were determined for the healthy mangroves and areas that measured 15.8 km?, Using a 0.5 LAT the remaining dwarf and poor condition man- interval up to 3.5, it is clear from the mapped D springerEnviron Monit Assess (2009) 157:137=149 Fig. 4. A comparative view of Isla La Palma mangroves from 1970 aerial photographs (a), an enhanced false color composite (NIR, R, G) of QuickBird data (b) and an estimated LAI that the Isla La Palma and sur rounding adjacent areas depict distinct mangrove zonation (Figs. 4 and 5). Discussion and conclusions From the results of this investigation it is sug- gested that the combined use of the AccuPAR LP- 80 and high resolution optical satellite data can provide resource managers with a quick and effec tive method for determining the current condition of their mangroves based on mapped estimated. LAI, With regards to the statistical significance of the linear regression models developed between these two sources of sensor data, similar levels of success were reported using another hand held. device, a LAI-2000 PCA, and comparable quality optical satellite data (Kovaes et al. 2004). How- estimated LAI map based on an NDVI linear model (©), For visualization, 0.5 intervals are shown for the estimated LAL ever, given that this previous study focused on a very different mangrove environment, dominated by red and white mangrove, itis difficult to deter mine which approach could be considered more accurate for rapid mapping of mangrove LAI, Nonetheless, with regards to the in situ devices employed, there are some differences that are worth considering, For example, in comparison to the LAT-2000, the AccuPAR LP-80 is less ex- pensive and has a much larger storage capacity. Moreover, unlike the LAI-2000, if the batteries fail, it is not necessary to immediately replace them in order to save the recorded data, With regards to its use in a mangrove forest, the lighter weight of the LP-S0 and, most importantly, the potentially longer period of time provided for collecting LAI of this device may make it a more desirable option, Specifically, if diffuse radiation from homogeneous cloud cover is not available D springer146 Environ Monit Assess (2009) 157:1374149 Bs Bee [2-10 Jono Fig. 5 An enlarged view of the southem section of Isa La Palma shown as an enhanced color composite (NIR, R, G) of QuickBird data (a) and as an estimated LAT ‘map derived from an NDVI linear model shown at an 0.5 interval (b). Dead stands are in grey aolor during daylight hours itis recommended the LAI- 2000 PCA be employed at dusk and dawn. In tem perate regions, these periods of diffuse radiation can be quite lengthy, but in tropical areas dusk and. dawn may last for a very limited period of time, For example, Kovacs et al, (2004) reported being, limited in collecting LAI-2000 sensor data to one hour before sunrise and one hour after sunset. ‘The logistical difficulties associated with reaching or leaving these mangrove forests at such times D springer does create difficulties. In contrast, the AccuPAR, LP-80 is ideally suited for collecting LAL in direct, radiation and is recommended for use near solar noon. Consequently, for this study area, a larger window for data collection is available and the logistical arrangements are more suitable with this, later sensor. However, mangrove LAL collection with the LP-80 may be more difficult to use in the rainy season or for other tropical areas where persistent nonchomogeneous cloud cover is the nom, In addition, another consideration worth noting is the processing of LAT when using the LP-80 with a separate datalogger. In particular, it is recommended in the user guide that a sepa- rate datalogger with an attached PAR sensor, as used in this study, be employed when obtaining, above-canopy measurements of timber or tropical crops, Asa result, for the post-processing of these data, the user must develop a program or spread- sheet to input the readings from the LP-80 and the datalogger into the below and above canopy readings, respectively, of the inversion equation in order to recalculate the LAT value. The inversion formula is supplied in the user's manual but there is no indication as to how one can recalculate the new fractional beam element of this equation, Fortunately, it was possible to receive a macro of this parameter after contacting the manufacturer directly. Although LAL is becoming increasingly more prevalent as a key biophysical parameter to moni- tor in a variety of global ecological studies (Asner et al, 2003), records of mangrove LAL are still quite scarce in the literature. Given the records available it is apparent that in comparison to ter resirial forests, mangrove LAI values for even healthy stands are quite low, For typical Rhi- zophora mangle dominated stands, mean LAT can range from values of 5.7 (Araujo et al. 1997) and 44 (Sherman et al. 2003) to as low as 2.49 (Kovacs et al, 2005) and 1.8 (Lovelock et al 2005), For dwart variants of this particular species, mean LAI values of 3,0 (Araujo et al. 1997) and 0,65 (Lovelock et al. 2005) have been recorded, For mature R, apiculaa, a mean LAT value of 4,9 was recorded in Peninsular Malaysia (Clough et al. 1997) and, similarly, mean values of 4.9 and 3.3 recorded in Vietnam (Clough et al. 2000). For dominant stands of black mangrove, AvicenniaEnviron Monit Assess (2009) 157:137=149 17 germinans, mean LAL values of 32 and 2.7 were recorded by Ramsey and Jensen (1996) in two basin stands located in southeast Florida. In quite tall white mangrove dominated stands, Laguncu- aria racemosa, Sherman et al. (2003) recorded an LAI value of 4.6. In contrast, in shorter, but not dwarf, white mangrove canopies experiencing sig- nificant degradation Kovacs et al, (2005) recorded. amean LAI value of 0.85. With regards to mixed. stands consisting of the most common mangrove species of the Americas (je. red, black and white), mean LAI values of 42 (Sherman et al, 2003) and 3.96 (Green et al. 1997) have been reported, Given these reported values, only the mangroves, along the edges of the Isla La Palma and the sur- rounding mainland, having a mean LAI value of 4,66, could be deemed as quite healthy mangrove. However, these fringe and riverine mangrove ar= cas represent only 8% of the mapped mangrove area, Excluding the dead stands, the mean LAI value of 239 for the remaining dwarf and poor condition mangrove areas appears consistent with, the values presented for somewhat stressed man- grove areas, Unfortunately, it is most likely that these LAI values, including those recorded for the healthy stands, are much lower than what would have been expected prior to the canal, Moreover, even the dead areas identified (LAI = 0), which represent 40% of the mangrove area, have been underestimated. Some previous mangrove areas have been replaced by dwarf glasswort (Salicornia bigelovil) and/or saltwort (Batis maritime) which currently depict non-zero values of SR and NDVI and thus were excluded from the analysis. Given these results it would indicate that, con- trary to others (Berlanga-Robles and Ruiz-Luna 2002; Fuente and Carrera 2005), that this northem portion of the Teacapin-Agua Brava-Las Ha- Ciendas estuarine-mangrove system, as with the southem section, has experienced considerable degradation, From the spatial pattems mapped using the estimated LAI (Figs, 4 and 5) it would also suggest that the hydrological modifications, following the opening of the canal, could have resulted in this extent of degradation, Most obvi- ous, is the fact that only the mangroves adjacent to edge of the main estuary or along the various tidal channels that penetrate more inland depicted relatively healthy LAI values. In contrast, the mangroves further inland, which may no longer be receiving ample amounts of natural flushing, show much lower LAI values with some areas, as shown in the older aerial photos, having been converted completely to dead stands, saltpans or saltwort/glasswort dominated areas. Although the results do suggest a formidable link to the canal it is difficult to confirm in the absence of histor= ical field records. Also of interest is the study from Pool et al. (1977), who, in their compari son with several other mangrove locations in the Americas (Florida, Puerto Rico, Costa Rica), re= corded mangroves from the particular island as one of the largest, with white mangrove as the dominant species. In our study, only a handful of relatively small DBH white mangroves were encountered in a few of the healthy mangrove ar- cas visited, However, exceptionally large diameter stems of white dead mangrove were also sporadi- cally encountered in these fringe areas, In conclusion, based on the results it is sug- gested that the use of the AccuPAR LP-80 with, very high resolution optical satellite data may provide resource managers with a very rapid and. effective method, based on mapped estimated, for determining the condition of their remaining man- groves, The instrument is relatively inexpensive and very simple to use in the field. However, it is recommended that the above canopy readings be collected separately using a datalogger with a PAR sensor in an open area and these data be later post-processed with the below canopy read- ings to provide more accurate in situ LA mea surements, One additional instrument required is a sub-meter GPS unit for recording the loca tion of the AccuPAR LP-80 measurements and, if need be, for collection ground control points, in geometrically correcting the optical data. With. regards to this study site, the results indicate that, in addition to the southem section of this system, these mangroves have experienced considerable degradation, However, unlike the southern sec- tion, the impacts in this area are strongly related to the distance from the edge of the tidal channels, and the main estuaries that feed in to them. Given the condition of these mangroves as well as the alarming rates of mangrove forest loss in Mexico and elsewhere in the tropics it is essential that these systems be urgently, but properly, mapped. D springerus Admowledgements The authors would like to thank Dr, Gaylon Campbell of Decazon Devices for providing ‘the beam fractional macro and Lance P. Aspden for his assistance in the field data collection process. This study was funded by a. grant (24949606) provided to John Kovacs from the Natural Sciences and Engineering Research Council of Canada, References Alonso-Perez, F., Ruiz-Luna, A. Tumer, J., Berlanga~ Robles, C. A., & Mitchelson-Jacob, G. (2008). Land ‘cover changes and impact of shrimp aquaculture on the landscape in the Ceuta coastal lagoon system, Sinaloa, Mexico. Ocean and Coastal Managanent, 45, 583-600, doi (0.1016/S0964-5691 ((B)OX36-X. Araujo, R. J., Jamarillo, J. C., & Snedaker, S. C. (1997). LAT and leaf size differences in twored mangrove for- st types in South Florida. Bulletin of Marine Science, 0, 643-647. Asner, G. P., Braswell, B. H., Schimel, D. S. & Wessman, CA. (1998). Boological rescarch needs from multian= ‘le remote sensing data. Ranote Sensing of Erviron= ‘ments 63, 155-165, doi|0. 10160342577 )0139, Asner, G, Pr Scurlock, Je M, On & Hicke, J. A, (2008). Global synthesis of leaf area index observations: Im- plications for ecological and remote sensing, stud- ies, Global Foology and Biogeography, 12, 191-206, oi 10. 1046). 1466-822X. 203,006.25. Beland. M., Goitas Ks, Bonn, F & Pham, T. Ts He (2006), Assessment of lanchaover changes related to shrimp aquaculture using remote sensing data: A case study in the Giao Thuy District, Vietnam, In- temational Journal of Ranote Sensing, 27, 1491-1510, doi: 10.1081701431 160500406888, Berlanga-Robles, C, A., & Ruiz-Luna, A. 2002).Land use ‘mapping and change detection in the costal zane of northwest Mexico using remote sensing techniques. Joumal of Coastal Research, 18,514-52, Brondizio, E., Moran, E., Mausel, P., & Wu, Y. (1996). Land cover in the Amazon estuary: Linking of the thematic mapper with botanical and historical data. Photogrammetric Engineeringand Remote Sensing, 2, 921-909, Clough, B. F., Ong, J. E, & Gong, W. K. (1997). Estimating eaf area index and photosynthetic produc- tion in canopies of the mangrove Rhizophora apic« tulata, Marine Ecology Progress Series, 159, 285-202. oi: 102354meps159I85, Cough, B. F., Tan, D. T., Phuong, D. X., & Buu, D. C. (2000). Canopy leaf area index and litter fallin stands ofthe mangrove Rhizophora apiculata of different age in the Mekong Delta, Vietnam, Aquatic Botany, 66, 311-320, dois0.1016/S0204-3770( 0900819, Dahdouh-Guetes, F., Van Hil, E., Chan, J. C-W., Jayatissa, LP, & Koedam, N. (2005). Qualitas tive distinction of congeneric and introgressive man= grove species in mixed patchy forest assemblages D springer using high spatial resolution remotely sensed imagery (IKONOS), Systematics and Biodiversin, 2, 113-119, oi: 10.1017/S1477200004001422, Flores-Verdugo, F. J. (1986). Ecoloafa de los manglares Yy perfil de comunidades en los sistemas lagunares de Agua Brava y Marismas Nacionales, Nayarit. ‘Technical Report, Consejo Nacional de Ciencia y Tecnologia (Clave: PCECBNAAZ2068, Mexico City, Mexico). Flores-Verdhgo, F. J., Gonzalez-Farias, F., Blanco-Correa, M., & Nunez-Pasten, A. (1997). The Teacapsin-Agua Brava Marismas Nacionales mangrove eoogystem on the Pacific coast of Mexioo. In B. Kjerfve, L. Dnude, & E. HS. Diop (Eds), Mangrove ecosystem stud: ies in Latin America and Africa (pp. 2546). Paris: UNESCO, Fuente, G,, & Carera, E, (2005). Cambio de Uso del Suelo en la Zona Costera del Estado de Sinaloa. Final Re- port to United States Forest Service (Grant No. 03DG- 11122762-157, Mexico: Ducks Unlimited de Mexico, AC., Gara Gattia,N. Gao, J. (1998). A hybrid method toward accurate mapping ‘of mangroves ina marginal habitat from SPOT multi- spectral data, International Joumal of Remote Sensing, 19, 1887=1899, doi (1080001431 1698215045, Green, E. P, Clark, C. D., Mumby, P. J., Exwatls, A. J. & Ellis, A. C. (1998). Remote sensing techniques for mangrove mapping. International Joumal of Remote Sensing, 19, 925-956, doi 0.1080001431 1698215801, Green, E. P., Mumby, P. J., Alasdair, E. J}, Clark, C. Da & Ellis A. C, (1997). Estimating, leaf area index of ‘mangroves from satelite data, Aquatic Botany, 58, I= 19, doi:l1O1G/S(BO+37 707X032. Herndnder-Comeja, R., Koedam, No Ruiz Luna, Aw Troell, Mz & Dahdoub-Guebas, F. (2005). Remote sensing and ethnobotanical assessment of the man- grove forest changes in the Navachiste-San Tznacio- Macapule lagoon complex, Sinaloa, Mexico, Eoology cand Society, 10,16. Hyer, E, Jp & Goetz, S.J, 2004), Comparison and sensi~ tivity analysis of instruments and radiometric methods for LAI estimation: Assessments from a boreal forest site. Agriculural and Forest Meteorology, 122, 157 174, dois0.1016).agrformet. 08.09.0153, Jensen, J R. (2007). Remote sensing of he environment: An ‘earth resource perspective. Upper Sacklle River, NI Pearson Echucation, Inc. Jensen, J. R. Lin, H., Yang, Y., Ramsey, E., Davis, B. A. ‘& Thoemke, C. W, (1991). The measurement of mane grove characteristics in Southwest Hlorida using SPOT multispectral data, Geoaarto Intemational, 2, 1321. Kovacs, J. M. (1999). Assessing mangrove se at the local scale. Landscape and Urban Plarning, 43, 201-208. doi: 10.1016/S0169-2046(98)00106-6, Kovacs, J. M. 2000). Perceptions of environmental change ina tropical coastal wetland. Land Degradation & Development, 11, 209-220, dei0.100D/1099-145X (200005106)! 13-20%: AID-LDR378>3.0.C02-Y. Kovacs, J. Mz Flores-Verdhigo, F., Wang, J, & Aspden, LP, (00), Estimating leaf area index of a de= graded mangrove forest using high spatial reso-Environ Monit Assess (2009) 157:137=149 149 lution satellite data, Aquatic Bony, 80, 12. oi:10.1016/j.aquabot.200405.001, Kovacs, J:M., Wang, Ju.& Blanco-Correa,M, (2001). Map= ing manerove disturbances using mult-date Landsat TM imagery. Envirormenal Management, 27, 76- 776. Goi:l0.1007/S000670010185. Kovacs, J. M. Wang, J., & Flores-Verdugo, F. (2005). Mapping mangrove leaf area index at the species level using IKONOS and LAF200 sensors. Es marine, Coastal and Shelf Science, 62, 377-384, doi:10.1016/j.ccss.2004,00.007, Lovelock, C. E. Feller, LC, McKee, K. L., & Thompson, R. (2005). Variation in mangrove forest structure and sediment characteristics in Bocas del Toro, Panama. Caribbean Journal of Science, 41 A56—464, Lugo, A.E., & Snexiaker, S.C. (1974). The ecology of man= groves. Annual Review of Eaology and Systematics, 5, 39.64, doi (0.1 146/annurees05.110174.000351. Mather, P. M. (1987). Computer processing of remotety= sensed images: An introduction. Toronto: Wiley. Muttitanon, W., & Tripathi, N. K. (2005). Land use/land cover changes in the’ coastal zone of Ban Don Bay, Thailand using Landsat TM. data. Jnterna= tional Joumal of Remote Sensing, 26, 231-2223. ddoi:10.10817014311605 12331336610, Pierce, L. L, & Running, S. W. (1988). Rapid estima tion of coniferous forest leaf area index using_a Portable integrating radiometer, Eaology; 69, 1762 1767. doicl 237941154, + Ex & Jensen, J. R, (1996), Remote sensing of mangrove wetlands: Relating canopy spectra to site~ specific data, Photogranmenic Engineering and Re- ‘mote Sensing, 62, 959-548, Pook Ds Ju Snodaker, S. Cx & Lugo, A. E. (1977). ‘Structure of mangrove forests in Florida, PuertoRioo, Mexico and Casta Rica, Biotropica, 9, 195-212. doi: 025072387581. Sherman, R, E. Fahey, T, J», & Martinez, P, (2003). Spa- ‘tial pattems of biomass and aboveground net primary Ram productivity in a mangrove ecosystem in the Demini- can Republic, Eoosystems (New York, N.Y.) 6, 3 38, doi:lOLO7SIO21N01I91-8, ‘Thu, P. M., & Populus, J, (2007), Status and changes of ‘mangrove forest in Mekong Delta: Case study in Tra ‘Vinh, Vietnam, Esmarine, Coastal and Shelf Science, 71, 98-109, doi:10.10164}.2cs52006.08.007. Tong, P. HS, Auda, Y., Populus, J, Aizpuru, M., Al Fabshi, A., & Bksco, F. 200). Assessment from space of mangrove evolution in the Mekong Delta in relation to extensive shrimp farming, Inter- national Journal of Remote Sensing, 25, 4795-4812. doi: 10.1080/01431 16041233 1270858, Valiela, 1., Boen, J. L., & York, J. K. (2001). Mangrove forests: One ‘of the world’s threatened major tropical environments. Bioscience, 51, S07ASIS. doi: 10.1641/0006-3568(2001)051[807: MFOOTW 20. co2. ‘Wang, L, Sousa, W. P., & Gong, P. (2004a). Integra. tion of object-based and pixel-based classification for ‘mapping mangroves with IKONOS imagery. Inter- national Journal of Remote Sensing, 25, 5655-5668. doi: 10.1080/01431 1600331291215, ‘Wang, L., Sousa, W. P., Geng, P., & Biging, G.S. 2004). Camparison of IKONOS and QuickBird images for mapping mangrove species cn the Caribbean coast of Panama, Remote Sensing of Environment, 91,4324, doi: 10.1016/.rse 200404005, Webb, EL, Evangelista, M. A., & Robinson, J. A. (2000). Digital land-use classification using’ space= shuttle-acquired orbital photographs: A. quantitative ‘comparison with Landsat TM imagery of a coastal en- vironment, Chanthaburi, Thailand, Photogrammetric Engineering and Ranote Sensing, 66, 1430-1449, White, J. D, Running, S, W., Ramakrishna, N., Keane, R, Ey & Ryan, K C, (1997), Measurement and remote sensing of LAT in Rocky Mountain montane ecosys+ ‘tems, Canadian Journal of Forest Research, 27, 1714 YTD. dH OM SV jft-27-1-1714, D springerReproduced with permission of the copyright owner. Further reproduction prohibited without permission.