Yes... nice to keep skills sharp! I got it to work... here's what I ended up with:
import arcpy
... import os
... #import codecs
... # parameters
... in_table = r'C:\Users\jwood\Documents\ArcGIS\Default.gdb\BGM_Chpt9'
... in_fc = r'C:\Users\jwood\Documents\ArcGIS\Default.gdb\OctDepths'
... out_folder = r'C:\Users\jwood\Documents\IUCN Red List\Shapefiles\Chpt9'
... polygon_code_field = "Code"
... #9
... # list all fields from the table
... all_fields = [f.name for f in arcpy.ListFields(in_table)]
... #12
... # loop through all rows in table
... with arcpy.da.SearchCursor(in_table, '*') as sc:
... for row in sc:
... #16
... item = dict(zip(all_fields, row)) # represent row as a dictionary
... #18
... species_name = item.get('BINOMIAL', 'Unknown')
... species_number = item.get('SpeciesNum', 0)
... iucn_num = item.get('IUCN_number', 'Unknown')
... origin = item.get('ORIGIN', 'Unknown')
... taxon_id = item.get('TAXON_ID', 'Unknown')
... compiler = item.get('COMPILER', 'Unknown')
... presence = item.get('PRESENCE', 'Unknown')
... citation = item.get('Citation', 'Unknown')
... seasonal = item.get('SEASONAL', 'Unknown')
... dist_comm = item.get('DIST_COMM', 'Unknown')
... year_ = item.get('YEAR_', 'Unknown')
... island_ = item.get('Island_', 'Unknown')
... source_ = item.get('Source_', 'Unknown')
... subPop = item.get('SUBPOP', 'Unknown')
... legend= item.get('LEGEND', 'Unknown')
... subSpec = item.get('SUBSPECIES', 'Unknown')
... chaptNum = item.get('ChaptNum', 'Unknown')
... tax_comm = item.get('TAX_COMM', 'Unknown')
... BGM_url = item.get('URL', 'Unknown')
... #38
...
... # get list of polygon codes that should be selected
... polygons_to_select = []
... for field_name, value in item.iteritems():
... field_name_upper = field_name.upper() # to make comparison case insensitive
... if field_name_upper and value == 1:
... polygon_code = field_name_upper
... polygons_to_select.append(polygon_code)
... #47
... if polygons_to_select:
... # create a string like "'A1','B2',..."
... codes = ",".join(["'" + str(pg) + "'" for pg in polygons_to_select])
... # build an SQL expression
... sql = '"%s" IN (%s)' % (polygon_code_field, codes)
... #53
... #print codes
... #print polygon_code_field
... #56
... # export as a new shapefile with species_name in the file name
... out_fc = os.path.join(out_folder, species_name)
... out_fc = arcpy.analysis.Select(in_fc, out_fc, sql).getOutput(0)
... #60
... # add the column with species number as string
... arcpy.management.AddField(out_fc, "BINOMIAL", "TEXT")
... arcpy.CalculateField_management(out_fc, "BINOMIAL", "'" + str(species_name) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "SpeciesNum", "TEXT")
... arcpy.CalculateField_management(out_fc, "SpeciesNum", "'" + str(species_number) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "ORIGIN", "TEXT")
... arcpy.management.AddField(out_fc, "TAXON_ID", "TEXT") #67
... arcpy.CalculateField_management(out_fc, "TAXON_ID", "'" + str(taxon_id) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "COMPILER", "TEXT")
... arcpy.CalculateField_management(out_fc, "COMPILER", "'" + str(compiler) + "'", "PYTHON_9.3") #70
... arcpy.management.AddField(out_fc, "PRESENCE", "TEXT")
... arcpy.CalculateField_management(out_fc, "PRESENCE", "'" + str(presence) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "Citation", "TEXT")
... arcpy.CalculateField_management(out_fc, "Citation", "'" + str(citation) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "SEASONAL", "TEXT")
... arcpy.CalculateField_management(out_fc, "SEASONAL", "'" + str(seasonal) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "DIST_COMM", "TEXT")
... arcpy.CalculateField_management(out_fc, "DIST_COMM", "'" + str(dist_comm) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "YEAR_", "TEXT")
... arcpy.CalculateField_management(out_fc, "YEAR_", "'" + str(year_) + "'", "PYTHON_9.3") #80
... arcpy.management.AddField(out_fc, "Island", "TEXT")
... arcpy.CalculateField_management(out_fc, "Island", "'" + str(island_) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "SOURCE", "TEXT")
... arcpy.CalculateField_management(out_fc, "SOURCE", "'" + str(source_) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "SUBPOP", "TEXT")
... arcpy.CalculateField_management(out_fc, "SUBPOP", "'" + str(subPop) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "LEGEND", "TEXT")
... arcpy.CalculateField_management(out_fc, "LEGEND", "'" + str(legend) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "SUBSPECIES", "TEXT")
... arcpy.CalculateField_management(out_fc, "SUBSPECIES", "'" + str(subSpec) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "ChaptNum", "TEXT")
... arcpy.CalculateField_management(out_fc, "ChaptNum", "'" + str(chaptNum) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "TAX_COMM", "TEXT")
... arcpy.CalculateField_management(out_fc, "TAX_COMM", "'" + str(tax_comm) + "'", "PYTHON_9.3")
... arcpy.management.AddField(out_fc, "BGMUrl", "TEXT")
... arcpy.CalculateField_management(out_fc, "BGMUrl", "'" + str(BGM_url) + "'", "PYTHON_9.3")
...
... #print island
... #print source
... #arcpy.Dissolve_management(out_fc, out_fc_Dissolv, "SpeciesNum", "", "MULTI_PART", "DISSOLVE_LINES")
...
... else:
... # no 'A1,A2' column contains 1 for this species, skip this species
... pass
10.2 is solid to work with in terms of extending functionality of the application through automation with Python! (I am jealous). Just curious, why you are associating what seems like a "random" floating table with shapefiles - shouldn't those features be tied with their respectable attributes? If this were the case, it would be SO much easier to automate (less convoluted and complex of an analysis project), since you could utilize the Select Layer by Attribute etc. i.e. draw from the power of querying alone, using python, and the script would select the diff polygons from the shapefile based on this boolean condition you have set up. Search Cursors are very powerful but can get very tricky, as we see here I suppose.. - extreme patience is needed of course when debugging. PS, Filip's code looks really clean and easy to work with, hope that works out for you. As is expected though it will most likely break the first few times you try and run it on a different machine! Part of the fun though right?! 
